Is Social Security a Ponzi Scheme?

June 5, 2026 – Almost to the day, today, eight years ago, was my last day at work. One thing I always looked forward to in retirement was never having to pay those dreaded payroll taxes again. Alas, eight years into retirement, I’ve picked up a few side gigs to stay involved and now run my own small financial advisory business. Not only do I pay Social Security and Medicare taxes again, but I now pay the full 12.4% Social Security and 2.9% Medicare taxes, i.e., the employee and employer portions out of my pocket. Ouch! I had hoped I would never have to put money into that stupid Social Security Ponzi Scheme again. Oh, wait, what did I just say? I must have heard this somewhere, probably from Elon Musk on Joe Rogan’s show. He probably said this mostly for the shock value without thinking too much about the financial nuances.

Nevertheless, the Ponzi Scheme comparison got me thinking: While Social Security is certainly not a literal Ponzi Scheme, where some scam artist runs off with the money, and the investors lose all their funds, is Social Security a Ponzi Scheme, at least to a degree? Are the ways in which Social Security differs from a Ponzi Scheme really only distinctions without a difference? How much better could I have done if I had invested my personal payroll contributions into the stock market or some other financial asset portfolio? All interesting questions! Let’s take a look…

What is a Ponzi Scheme?

Charles Ponzi promised a very attractive investment scheme: give him money today, and he can double that within 90 days, a rate of return of about 1,500% annualized. He claimed that he found an arbitrage scheme involving so-called international reply coupons (prepaid stamps for international mail) that would fetch this astronomical return target. Of course, none of the money he received from investors went to his alleged investment. Instead, he funded his lavish lifestyle and paid early investors with money coming in from new investors. The scheme’s popularity outpaced his cash-flow needs, and he was able to continue paying early investors from the new cash flows. Helping him keep the scheme alive was that many investors simply reinvested their “earnings.”

Eventually, of course, the system collapsed, and investors lost about $20m, which is over $300m in 2025 dollars. Charles Ponzi lived a very lavish lifestyle, and some early investors who were smart (lucky?) enough to withdraw their money from the system also benefited from the scheme. But all subsequent investors were left holding the bag. Luckily, the scheme was shut down early enough. There were some remaining assets for the investors to recover at least 30 cents on the dollar. Intriguingly, the largest Ponzi Scheme ever was Bernie Madoff’s hedge fund, where investors recovered over 90% of their initial investments. There, the court-appointed trustees clawed back large sums of ill-gotten gains that had been paid out to early investors. Of course, investors never recovered any of the imaginary and fraudulent paper gains.

Social Security has features similar to a Ponzi Scheme

The similarities between a Ponzi Scheme and Social Security are numerous. In Social Security, an initial generation of beneficiaries received above-average returns, in fact, as high as +infinity for the generation that got benefits without ever paying into the system. The politicians who passed the law got the benefits of pleasing their constituents, a.k.a., buying votes. Just like Charles Ponzi never invested anything in the international stamp scheme, Social Security is a pay-as-you-go system. Moreover, the “free money” and high internal rate of return (IRR) of the initial generation come at the cost of much lower IRR for future generations, more on that below.

Some folks criticizing Musk pointed out that Social Security has government backing, unlike the typical Ponzi Scheme. But government involvement is a two-edged sword. In fact, a government mandate makes any potential Ponzi Scheme worse, not better. With a privately run Ponzi Scheme, at least I have the option not to participate.

Moreover, the government’s backing behind Social Security is not 100% watertight. Unbeknownst to many, we have no legal ownership, no contractual right, and no property rights in the FICA taxes we pay over our lifetime. Theoretically, if the government wanted to take away our Social Security benefits, we’d have no legal recourse. The U.S. Supreme Court decided so in Flemming v. Nestor (1960). Of course, a widespread rug pull is highly unlikely because it’s politically unattractive – old folks are an important voting bloc – but there is nothing keeping the government from a gradual expropriation of your Social Security benefits. It has happened before, i.e., through increasing the age at which you can claim your full benefits. It can happen again, especially considering the current underfunding, i.e., once the Social Security Trust Fund is depleted, likely in 2033, the payroll tax revenues only cover about 77% to 81%, according to this 2025 Social Security Administration report.

The one big difference between a Ponzi Scheme and Social Security

The one major feature inherent in a Ponzi Scheme is that it will eventually collapse, causing a large group of investors to incur significant losses, up to the total loss of their investments. After that, a Ponzi Scheme has no more victims. That is very different from Social Security. That got me thinking: How could we have saved Charles Ponzi’s scheme? For example, the government could have stepped in and “recruited” another generation of investors. Take their “investments” and distribute the money among the Ponzi victims. The government no longer promises Ponzi-Style rates of return, but something much more manageable, say inflation plus the rate of population growth, and then successively forces ever-new rounds of investors to chip in and make whole the previous round of investors. None of the investments is actually put to productive use, but we have instead created a pay-as-you-go system in which one initial cohort (Charles Ponzi plus a few of his early investors) had a windfall profit, while all subsequent “investors” have a slightly positive, albeit below-market, return on their investments.

Of course, nobody would want to continue investing in that Scheme at such a low rate of return, but the government’s taxing authority could clearly make that happen. That sounds like Social Security to me! Just like in the payroll tax-funded Ponzi Scheme, Social Security avoids a total loss to any one generation by spreading a large loss over all future generations. That loss may be small enough for everybody to stomach, especially when the real return is still (slightly) positive, and we factor in other attractive features of Social Security, e.g., longevity insurance and care for widows and orphans. So, how high or low is the implicit return on your Social Security cash flows? To better understand the return expectations, let’s build a simple model to study this question, which brings me to the next section…

What is the expected return of Social Security? A simple multi-generation model of a Pay-As-You-Go system

Imagine we have an infinitely lived economy, and at any point in time, there are two generations present: young and old. Let’s assume initially that both generations are of equal size in every period. Assume one time period is as long as an entire generation (e.g., about 30 years). So, the young generation in period t will be the old generation in period t+1. Also, the older generation in period t will die and no longer be around in t+1, while a new young generation enters the landscape.

Let’s assume that in period T, the current president proposes a pay-as-you-go system that taxes each worker in the young generation and gives the proceeds to the current old generation. To sell this to the young generation, he tells them, “Don’t worry, when you’re old in period T+1, you will get a transfer from the then-young generation.” Of course, this new transfer program would be wildly popular with the initial old generation, for they get money for nothing. Subsequent generations receive a return equal to the growth in the total wage base, assuming the marginal tax on labor income remains constant. But the real growth rate of the wage base is likely much lower than what you’d get in the stock market!

For example, in the chart below, I plot the cumulative gain in the wage base compared to CPI inflation and the S&P 500 Total Return index (dividends reinvested). Note that the wage data series is not per-person wages but total wage income across all workers, to gauge the potential return of an expected pay-as-you-go system according to the overlapping generations model. Quite intriguingly, the total wage base increased at a pretty attractive rate: 6.07% nominal and 2.30% if adjusted for inflation. 4.47% and 0.74%, for nominal and real growth, respectively, over the last 30 years.

Summary so far: The political appeal of a Social Security-style transfer program is that the initial generation gets a magnificent windfall, i.e., they get benefits without ever paying anything into the system. If the system had just been a one-time affair, this would have been a zero-sum game, i.e., the younger generation footing the bill would bear 100% of the burden and essentially suffer a 100% loss of their contributions. Just like in a Ponzi Scheme. However, by keeping the system alive for many more, even infinitely many generations, we spread the cost of the free gift to the initial generation over sufficiently many other people. Spreading the cost wide enough can make the loss small enough that each future generation may not even notice. But how small or big is the cost to the subsequent generations? Is the internal rate of return (IRR) of someone’s Social Security contributions and benefits much lower than expected financial market returns? This brings me to the next section…

Social Security Return Sample Calculations

Let’s consider a 67-year-old retiree who begins receiving benefits at their regular retirement age. What’s the value of the Social Security benefits? Let’s perform an actuarial exercise.

Let’s assume that several fresh retirees claim Social Security today in 2026. They look back at their earnings and payroll contribution history, calculate their benefits, and plug the Average Indexed Monthly Earnings (AIME) into the Social Security benefits formula; please see the chart below. Note the two bend points where the slope of the benefits function shifts from 0.9 to 0.32 and then to 0.15. This concave shape of the benefits function creates redistribution: retirees with higher incomes, while receiving larger total benefits, have a lower replacement ratio (benefit divided by AIME).

Side comment: Who pays the employer portion of payroll taxes? I assume that the full cost of Social Security payroll taxes (currently 12.4%, though this rate has varied between 10.4 and 12.4% over the last 45 years) is borne by the employee. Why not include only the employee portion? Simple, your employer is not a charity. They will lower your salary offer to account for that cost, of course. Similarly, without Social Security, you would likely receive a pay raise.

I consider six different work histories:

• Retiree 1: 45 years of earnings from 1981 to 2025, earning the maximum taxable earnings in each year. This is the worst-case scenario from an internal rate of return perspective because the Average Indexed Monthly Earnings (AIME) accounts for only the top 35 years of indexed earnings. In addition, the AIME is the maximum level, so you’re far into the 0.15% bracket, where the benefit as a percentage of your contributions is minimal.
• Retiree 2: 35 years of earnings from 1991 to 2025, earning the maximum taxable earnings in each year. This is slightly better than the scenario above because you avoid the ten wasted years that add essentially zero to your AIME as in the worst-case scenario. But you’re still far into the 15% bracket of the AIME vs. benefits chart. The benefit as a percentage of your average earnings will be very low.
• Retiree 3: Same as #2, but the earnings are exactly half of the maximum taxable amount every year. Thanks to the concavity of the benefits function, the monthly benefit is only about 28% lower ($3,037 vs. $4,211) despite paying in exactly 50% than retiree 2.
• Retiree 4: Same as #2, but now earnings are only 20% of retiree 2. But the benefits are significantly better than 0.2x of retiree 2. Despite paying into the system 80% less than retiree 2, the benefits are only about 60% lower.
• Retiree 5: This should be the best possible Social Security IRR, i.e., the shortest possible contribution history (you need 10 years to qualify), the latest possible history, i.e., the 10 years right before retirement, and the contributions are low enough to still land you in the 90% slope portion of the benefits.
• Retiree 6: I also want to include the typical FIRE career. Someone who claims benefits in 2026, but was already FIRE’d for more than 20 years after a 15-year high-paying career, making the maximum taxable amount every year between ages 31 and 45.

Since the value of Social Security benefits depends on your life expectancy, I also calculate the IRR for four different types of individuals. I use my actuarial sheet; see Part 56 of my SWR series for more details, and I use a real discount rate of 2.5% per annum.

• A: A male with below-average life expectancy. A life expectancy of about 14.8 years. The actuarial calculations imply that a $1 monthly Social Security benefit, adjusted for CPI, has a discounted expected value of $143.08.
• B: A male with average life expectancy. A life expectancy of about 16.7 years. A $1 monthly Social Security benefit, adjusted for CPI, has a discounted expected value of $157.74.
• C: A female with average life expectancy. A life expectancy of about 19.1 years. A $1 monthly Social Security benefit, adjusted for CPI, has a discounted expected value of $176.48.
• D: A female with above-average life expectancy. A life expectancy of about 21.8 years. A $1 monthly Social Security benefit, adjusted for CPI, has a discounted expected value of $195.60.
• What about men with above-average health or women with below-average health? Simply use individuals C and B, respectively.

Let’s look at the results; please see the summary table below:

• In the absolute best-case scenario for Social Security, Retiree 5, not even the roaring bull market over the last 10 years would have created an equity portfolio large enough to compete with the Social Security expected benefits: $66,709 for the equity portfolio, compared to $104,332 to $142,628 in Social Security benefits. All other retirees would have done significantly better investing their contributions in the stock market.
• If equity investments were not your thing, your Social Security cash flow series could have outperformed investments in 3-month T-bills in all but the most disadvantaged retirees: Retiree 1A and 1B, i.e., with the longest contribution history and the shortest life expectancy: males with average or below-average life expectancy.
• Investing in 10-year US Treasury bonds (closely approximated by, say, the iShares ETF, ticker IEF) would have done slightly better than T-bills. But still, most retirees would have done better with their Social Security “investment.” Everyone except Retiree 1, Retiree 2A and 2B, and 6A.
• The Social Security IRRs are also interesting:
  • Of course, retiree 1 did quite poorly, with an IRR that didn’t even keep pace with the average inflation rate of 3% over those 45 years. Retiree 2 also just gets roughly a zero real return, though women do a little bit better. But the low returns for retirees 1 and 2 are by design: Social Security has an element of redistribution that takes money from extremely high earners and subsidizes benefits for low earners and/or people with a shorter earnings history.
  • Retirees 3 and 6 got pretty decent returns between about 4% and 6%, depending on exact parameters. That’s pretty close to the “theoretical” Social Security return, calculated as the average annual growth in the wage base.
  • Retiree 4 has significantly below-average earnings and thus receives very attractive nominal returns of 5.83% to 7.4%.
  • Retiree 5 has the most impressive returns, all in double digits, ranging from 19.94% to 25.59%, depending on gender and health status. Much higher than the equity market.
  • This all makes sense: a more “average” earnings history should give you the average Social Security expected IRR. Higher earners do worse, and lower earners do better due to the redistribution.

Limitations

I’m aware that I took some shortcuts in my calculations. On the one hand, by basing my calculation on retirees who are 67 years old today, I ignored the possibility of someone dying before reaching that age. So, that would lower the attractiveness and the IRR of Social Security because investors who die before age 67 could have bequeathed assets to their heirs, while their Social Security benefits are forever lost. On the other hand, Social Security has some additional built-in benefits: disability benefits while working, survivor benefits, and the possibility for spouses with low or no earnings to qualify for 50% of the higher-paying spouse’s benefits. All of this would have been quite complicated to model in a short blog post, so I cross my fingers that the two effects roughly cancel out.

I also ignore taxes. Including the tax effect will be detrimental to both Social Security (because up to 85% of your benefits are taxable in retirement) and the investor. Taxes could create a drag both during the accumulation phase (dividends and interest income are taxed along the way) and in retirement. Though I suspect the tax impact on the investor is often lower because the investor pays tax only on gains, not on the cost basis. Only when investing heavily in equities over many decades, with capital gains making up 85%+ of the final portfolio, would the tax be as impactful as in the Social Security case. However, even in that case, the tax on capital gains is still lower than the ordinary income tax on Social Security.

Social Security also has a feature that’s not easily replicable with financial instruments. So, even if you had, say, $479,073 (retiree 3.B), you couldn’t so easily transform it into a $3,037 monthly benefit, adjusted for CPI inflation, even though the table above spits out the numbers for retiree 3 and an average-healthy male. That’s because there appears to be no functioning market for CPI-adjusted annuities. You will find the most competitive prices for an SPIA (Single-Premium Immediate Annuity), which is unfortunately, not CPI-adjusted. But the SPIA could still get you close: I got an SPIA quote for a 67-year-old with $479,073, and he could get a monthly benefit of $3,441. The retiree could invest the extra $3,441-$3,037=$404 each month to fund future cost-of-living adjustments, though this would not be a truly safe hedge against inflation and longevity. Also, forget about applying the 4% Rule! The $479,073 portfolio in this example would only generate just under $1,600 in monthly benefits. Social Security generates larger benefits than the 4% Rule because of the mortality risk: if you die one month after Social Security starts, your money is gone, while 4% Rule investors can give their leftover portfolio to their heirs.

What about the prospect of benefit cuts? After all, current Social Security payroll tax revenue will cover only about 80% of the promised benefits. A full 20% benefit reduction across the board would certainly reduce your Social Security IRR. We can gauge this by examining the IRR deterioration when moving from the expected discounted benefits of Individual D (Female with above-average health) to those of Individual B (Male with average health), with the latter being roughly 20% lower. That reduced the IRR by well over a percentage point. So, even if retirees were to bear the entire cost, Social Security wouldn’t become a Ponzi Scheme. The IRR will merely be slightly lower. And keep in mind that not all the costs of the future Social Security reform will be borne by the current retirees. Any benefit reduction will likely be phased in slowly, probably through increases in the Full Retirement Age and the payroll tax. So, my Social Security IRR will likely not decrease by that full percentage point. I’m crossing my fingers!

What if AI takes away all the jobs? Well, that would be a concern for a pay-as-you-go retirement system because the “returns” are tied to the growth of the wage base. Being the eternal optimist, I predict that AI will cause productivity gains and some job losses, but will have an overall positive effect on all economic variables, i.e., productivity, GDP growth, earnings, etc., just like all other productivity breakthroughs before AI.

Conclusion

True, Social Security shares some features with a Ponzi Scheme. It’s also disconcerting that future governments could mess with our benefits without beneficiaries having any legal recourse. But after beating up on Social Security, I want to end this post in a conciliatory tone. Sure, I could have done much better investing my Social Security contributions in the stock market. But first, that wasn’t in the cards; we, as a society, can’t repudiate the obligations to our current retirees. And second, in absolute terms, the implicit returns on my Social Security contributions (even after accounting for both employer and employee costs) are not that bad. Especially retirees with lower incomes and/or without a full 35-year work history (the FIRE crowd!), who still fall into the 32% bracket of the benefit formula, got pretty decent returns. Less than equity returns, but comparable and in many cases even superior to nominal bond or money market returns. Heck, over the last 20 years, your Social Security contributions might have outperformed Small-Cap Value stocks, but let’s not go there. So, let’s stop beating up Social Security. By design, as a Pay-as-you-go system, it can never perform as well as the stock market. But its IRR is still high enough to make Social Security a useful tool.

Thanks for stopping by today. I look forward to your comments and suggestions!

Title Picture: WordPress AI picture generator