Americans have long been led to believe that Social Security is a rock-solid system. But from very early on, Social Security had to be regularly modified to remain financially viable. At its inception, the system had a combined employer-employee tax rate of 2 percent levied against the first $3,000 of earnings. That held until 1950, after which the payroll tax rate and the earnings that are taxable had to be continually adjusted upwards. In a statement that only applies through 2009, the system's administrators admit to this: "Since 1937, there have been 11 years in which benefits paid out exceeded income and so the assets of the Trust Funds had to be spent to make up the difference."

By the 1970s and early 1980s those shortfalls were becoming an annual problem. To put Social Security back on a self-financing, pay-as-you-go basis, Congress enacted the Social Security Amendments of 1983 and raised the combined employer-employee tax rate in stages from 10.8 percent to 12.4 percent, where it's remained since 1990.

That 1.6 percent rate hike may not seem like much, but when combined with the annual hike in the earnings that are taxable, it helped pay for some hefty boosts in benefits as well as generate big surpluses. Those surpluses were put into the so-called "trust fund" held by the Social Security Administration, which are now reportedly worth all of $2.7 trillion. Inasmuch as the OASI "trust fund" in 1983 contained less than $20B, could the increase in payroll tax revenue really have paid for those big benefit hikes and have produced annual surpluses that added up to $2.7T?

One way to confirm that the surpluses produced by the payroll tax had in fact amounted to $2.7T since 1983 is to add up yearly payroll tax revenue as well as benefit outgo and then subtract the latter sum from the former. It'd be nice if such data were in a spreadsheet and one could just sum it; take about 5 seconds. But I hadn't been able to find such. So to ascertain just how much of the $2.7T surplus is actually surplus, I went to Table VI.A1. --- Operations of the OASI Trust Fund, Calendar Years 1937-2012 and tediously added up "Net payroll tax contributions" as well as "Benefit payments" for 1984 through 2012 on my trusty calculator.

But you don't need to verify my sums by laboriously adding two columns of 29 numbers, because I finally found the Excel spreadsheet and it comports with my hand calculations. (I found the spreadsheet here.) Anyway, I got $10,874B for net payroll tax contributions and $10,111B for benefit payments, which means that the Social Security portion of the payroll tax over those 29 years produced a total surplus of $763B.

So less than one third of what Sen. Elizabeth Warren calls a "$2.7 trillion surplus" is actual surplus; two thirds of it is something else. What is it?

"What's In the Social Security Trust Funds" by Charles Blahous shows that it's mainly interest. Mr. Blahous seems to be a pretty interesting guy, having received a Ph D "in computational quantum chemistry from the University of California at Berkeley." But what might give his article an extra, added authority is that he is "a public trustee for Social Security and Medicare."

Blahous's article came out in December 2011 when Congress was considering extending the payroll tax holiday for another year. What concerned Blahous was Social Security's continuation as "a self-financed, earned benefit program," but what caught my attention were his two charts. The first of these, "Components of Social Security Trust Funds Balance, EOY 2011," seems consistent with my computations above. It shows "Previous surpluses of tax income over expenditures" as being 35 percent of the fund, and "Interest credits" as being 60 percent. (Where's Liz Warren's "$2.7 trillion surplus" now?)

The second chart shows the effect of extending the payroll tax holiday into 2012, and it shows a dramatic change. Blahous then predicts that if "extended indefinitely, by early 2015 the entire $3 trillion balance of the Social Security Trust Funds would be attributable to general revenue subsidies and interest credits."

That didn't compute; going through 28 years of surplus in three years just didn't seem right. Rereading the lead-in to the chart, I saw that Blahous was not just talking about extending the tax holiday, but of deepening it, too, in accordance with the "President's proposed policy." So I clicked on the links in the article and found that an earlier article by Blahous explained that the deepening would be due to enacting the American Jobs Act, and the math of Blahous's forecast then made sense. Thankfully, the payroll tax part of the American Jobs Act of 2011 wasn't enacted. However, the 2 percent payroll tax holiday was extended through 2012.

Although we're back from holiday, and the full 12.4 percent bite of the Social Security tax has been restored, the accumulated surpluses are still quickly disappearing, which will ultimately leave the "trust fund" with little but interest.

By my arithmetic, the Social Security portion of the payroll tax produced surpluses of $1.043T in the "pay-as-you-go era" from 1984 through 2009. But in 2010 Social Security went cash-flow negative, and is expected to remain so. Despite that, the "trust fund" posted gains in each of the years since 2009 because of interest.

In A Summary of the 2013 Annual Reports, the introductory Message to the Public reads: "The Trustees project that this cash-flow deficit will average about $75 billion between 2013 and 2018 before rising steeply." At that rate, the $763B credited to tax surpluses will be gone in about nine years, leaving mainly interest. And if the steep rise in cash-flow deficit after 2018 were steep enough, we could eat through the remaining surplus even sooner, but the "trusts fund" would still have interest.

We see on page 2 (PDF 10)of the summary that for 2012 OASI retirement benefits were $637.9B while payroll taxes were $503.9B, a shortfall by my lights of $134B. Even so, the "trust fund" posted an $85.6B gain in 2012, all because of interest.

Regardless of how much of the "trust fund" is surplus and how much of it is interest, it all amounts to the same thing: promises. To keep those promises, the feds must raise new taxes, sell new bonds, and/or print new money. Any Social Security benefit paid for out of past surpluses has been paid for twice. Social Security needs to again be cash-flow positive; it needs to operate on a pay-as-you-go basis; it needs to pay benefits with revenue that comes entirely and directly out of the payroll tax.

Consider this: What if, rather than raising the payroll tax rate by 1.6 percent as we did thirty years ago, Congress had instead raised it only enough to pay benefits and raised income taxes enough to offset the lost payroll tax surpluses? Using a static analysis that assumes Congress could exactly calibrate rates to exactly accomplish that, this is what that alternate tax regime would have bought us: Social Security would have paid out the same in benefits, the publicly held debt (which is the real debt) would be exactly what it is today, and the feds would have hit taxpayers up for exactly the same amount of money. Everything that's real would have been the same.

What wouldn't have been the same is what's unreal: The total (i.e. phony) national debt would be $2.7T lower, the "trust fund" would contain about $20B instead of $2.7T, and, most importantly, we wouldn't have had to listen for thirty years to politicians tell us lies about the solvency of some mythical "trust fund."

Politicians tell us that the Social Security "trust fund" will last until 2033. But long before 2033, probably in less than nine years, the payroll tax surpluses will be gone and the "trust fund" will consist mostly of interest. To keep the "trust fund" alive, one wonders if the government will pay interest on that interest.

Jon N. Hall is a programmer/analyst from Kansas City.