Press "Enter" to skip to content

The Simple [yet counterintuitive] Economics of Social Security

Originally published at Autos and Economics on August 11, 2010
Social Security Projections, 80% Confidence Intervals

Pay-Go and the Trust Fund
Economic Structure
Most consumption in the United States is of services, not goods; of the rest, a significant proportion consists of time-sensitive goods, either because of their physical nature (food), because they reflect style (clothing) or because they are subject to aging and obsolescence (cars, cell phones, computers). As a result, as a first approximation goods and services must be consumed when they are produced; they cannot be stored.[1] That distinction is critical for understanding the economics of retirement (as opposed to the legal institutions and the politics of the Old Age, Survivors, and Disability Insurance or OASDI, to use the official name of “social security”).[2]
It is natural to think of social security as operating in a manner analogous to a household, and to think of the Trust Fund in that context. In this instance, that analogy is simply wrong. A household can reduce its consumption today in order to save and build up financial resources that can later be sold to permit greater future consumption. An entire economy cannot. Even in a purely private retirement system, retirees can consume more only if workers consume less. If workers are uninterested in saving, then retirees will find few buyers for their assets and won’t be able to consume much, even if they lived frugally while working in order to save more for retirement. Conversely, they might retire at a time when every worker wants to buy stocks and real estate, and find they can live very well; a couple of my colleagues retired just before the 2000 “” bubble broke, to find that their monthly income exceeded that while they were working. But with the mass of baby boomers retiring en masse, the latter scenario looks less likely, though attempts to quantify that effect don’t suggest that it will large in magnitude.[3]
What this means is that for social security to be economically solvent (as opposed to institutionally solvent), the government must tax and/or borrow commensurate with outlays. The amount needed at any given point in time is straightforward to calculate: the tax must equal (i) the “replacement rate” of social security (monthly pensions as a percentage of current workers’ taxable incomes), current about 38%, (ii) divided by the ratio of workers to retirees, which given unusually high unemployment in 2010 was about 2.9. So on a per worker basis, covering payments requires a tax rate of 38/2.9 = 13.1%. Since the current rate is 12.4% that leaves the system with a modest deficit, which will be reversed if employment picks up during 2010-11.[4]
As the baby boomers retire, the ratio of workers to retirees will with certainty decline. As a result, the implicit tax that social security represents will increase. The August 2009 Congressional Budget Office estimate is that estimate is that this will result in outflows exceeding revenues in 2017. The gap will increase in their baseline scenario to 1.3% of GDP until 2035, when the baby boomers will begin dying off, and will then shrink to about 0.8% of GDP. However, unless there is an increase in mortality rather than in longevity, the population will remain more aged than in 2010, and after about 2050 the gap will increase again, albeit at a very slow rate.
Now 1% of GDP may sound like a small number, and in an economic sense 1% of GDP is a small number. Of course with an economy of $15 trillion, 1% means $150 billion a year, small stuff compared to the Department of Defense. But those who oppose Social Security for one or another reason can manipulate that number. Social security has a legally mandated 75-year planning horizon; add up $150 billion a year for 75 years and you get a large number, by intent a frightening number; even discounted for compound interest, it is equivalent to a $3.7 trillion increase in national debt.[5] But when viewed on an annual basis, now and into the distant future, a modest shift in taxes — again, far less than the budget of the Department of Defense — would be enough to cover projected shortfalls.
Institutional Structure
In the early years — from 1937, when it was enacted, and throughout the 1940s — there would clearly be very few Social Security recipients, but it would not would not remain that way. So the decision was made to raise taxes proactively, and to try to “set aside” the excess by segregating it from the regular budget in a “trust fund”. The institutional hope was that other taxes would be adjusted to keep the “regular” budget in balance, so that the overall budget could be kept in balance as the number of retirees increased without needing to “enhance revenue” for that purpose. Now Congress in fact did behave on an “every ship on its own bottom” basis until the mid-1960s: the “regular” budget was kept in balance while social security accumulated surpluses. That’s no longer the case but is not economically relevant; after all money is fungible — one tax dollar looks like any other tax dollar. Indeed, recession-linked temporarily shortfalls aside, subsequent adjustments to the social security tax mean that almost 75 years later inflows continue to exceed outflows. That will change around 2017, depending on how many people retire early in the current recession, and on the rate of recovery of incomes of the lower nine-tenths of the US population.
But even in 2017 there will be no crisis, because the US government isn’t suddenly going to be broke, and in addition payments are legally tied to the Trust Fund. As long as there are (paper) assets remaining there, Social Security checks will continue to be cut each month without any need for action by Congress. That will change circa 2039, under current projections. At that point things start to change, because under the (by that point) century-old enabling legislation Social Security payments will be limited to the level of FICA (Social Security payroll tax) receipts. Absent change in the interim, total payments would fall in 2039 from roughly 6.0% of GDP based on the number of retirees to 4.9% of GDP based on anticipated same-year FICA receipts. That represents roughly a 20% drop in benefits. At some point the legislation will need to be amended. Given the political importance of social security recipients, it surely will be.
The alternative would be to allow expenditures – the size of social security checks – to fall. Retirees, current and future, paid taxes for decades, and planned on social security to supplement retirement. (Relative to Europe, and relative to pre-retirement incomes in the US, social security isn’t particularly generous; it was intended as a supplement. Conceptually, allowing payments to fall 20% would rob retirees of anticipated income and be every bit as much a tax increase as (say) bumping the Federal income tax or (the preference of most economists) levying a national sales tax. I personally would view that as unethical and because the impact would be on a narrow group, unfair.
Over most of its history social security has “paid” less in a financial sense than average private returns; given the insurance implicit in defined benefit pensions, it is much more neutral. Because they ignore the size of the insurance premium, which the social security trustees are not required to calculate, conservatives have argued that the system ought to be privatized as unfair to retirees. If they want to be intellectually consistent — I know, dream on — that implies that if the current system isn’t done away with, conservatives are effectively arguing that social security payments to retirees ought to be increased. So I am puzzled (well, disturbed) that the same people who are arguing that fairness mandates a privatized system are at the same time arguing that benefits must be cut because we can’t afford them.
I fear however that our political system is too fractured and places too much emphasis on short-run campaign strategies for us to address the structure of the social security system today rather than in 2037. Some of the crying wolf over “crisis” is from economists who observe the political dynamics and want to try to create a sense of urgency, not because they fail to understand the nature of retirement systems. I fear however that their strategy is counterproductive because it conveys the impression that we as a society can’t afford to maintain the social security system. That is simply not the case; the US remains and in 2037 will remain the richest large country in the developed world, and social security is not a big slice of the economy. Now if lessening the overall Federal deficit is important — and I believe it is — it may be politically necessary that everyone is seen as bearing some of the cost of adjustment. The basic ignorance of the fundamental difference between household and societal retirement however seems to fuel demagoguery that may lead to a disproportionate emphasis on this one program.
So . . . . if for argument’s sake, we assume that all of the social security shortfall were to be covered by higher social security taxes (FICA), how large of a change would this be? With current receipts of 4.8%-4.9% of GDP, “saving” social security in the long run will require boosting this at peak to 6.1% or by 1.3/4.8 = 27%. With the current rate of 12.40%, this would imply we would need a peak rate of 15.75%. While no one likes higher taxes, this is hardly a shift that is infeasible from an economic perspective. If there are problems with the sustainability of social security, then they are political. Other alternatives include increasing the tax base rather than the tax rate (payroll taxable for FICA purposes is only 38% of GDP) or cutting benefits (including by increasing the retirement age). Given that social security is but one contributor to the fiscal challenges facing the US, it may be politically convenient to consider it as a separate entity, but the fiscal pressures that stem from social security itself could be addressed by shifts in taxes and expenditures elsewhere in the budget. If we could go for a decade without getting involved in another war…sigh.
That returns the discussion to the relevance of the Trust Fund: does it have any economic significance? (It clearly is significant from a legal-institutional perspective and, because of the “drop-dead” nature of changes triggered circa 2039, it has large future political implications.) The answer depends on whether (i) it served to increase national saving (via a smaller budget deficit) and in turn whether (ii) changes in national saving are linked to investment in productive resources.
Until 1965, the Trust Fund was kept separate from the overall budget, and so until that point it may have had an impact on national saving, because it did not serve to reduce the operating deficit with which Congress dealt. Thereafter, however, the “off budget” distinction has not been substantive. The budget record suggests that, to the extent that Congress pays attention to balancing budgets, it has done so by looking at the entire budget, and not the budget net of social security surpluses. So the evidence is that the Trust Fund has not served to increase national saving. Furthermore, as a share of GDP the US has run large balance of payments deficits since roughly 1980, or in other words, we have been able to borrow from the rest of the world. While the situation is not uniform across countries, that means that in the case of the US, there is no longer a strong link between national savings and investment. At some points in the past 3 decades US real interest rates have been higher than in Europe and Japan, so there is evidence of some link, but fundamentally the magnitude of national savings has not been a binding constraint on investment. So the answer to the question(s) of whether the Trust Fund has been economically significant are (i) “no” and (with qualifications) (ii) “no.”


  1. Small “open” economies — Belgium — have more leeway because imports (services included) are a far greater share of consumption; foreign assets can accumulated now and sold as the population ages. That’s not applicable in the US context because we are too large and physically remote to import much, particularly of services. In any case we as a nation are substantial net debtors to the rest of the world: it’s too late to use the international margin of adjustment. If anything it could work against us if we are forced to “rebalance” (export more, net) because the rest of the world wants to sell their US assets.
  2. I ignore the survivor and disability insurance components here, because they are small relative to the purely retirement component.
  3. Takats, Elod. “Ageing and Asset Prices.” Basel: Bank for International Settlements, August 2010, BIS Working Paper No. 318. This paper finds an effect (unlike previous papers that were relatively optimistic) that the author characterizes as “headwind not meltdown.” Note too that this up-and-down nature of private assets values should be a reminder that social security includes an insurance component that is lacking in private alternatives, indeed that is impossible to purchase on a private basis.
  4. One fallacy that is widely repeated is that Social Security is a Ponzi scheme. However, Ponzi schemes blow up because you need ever-increasing numbers of “marks” for the scam to operate, exponentially increasing. They cannot be sustained except in the short term, and there is no way that everyone can get their money out. That is not true of social security. It is fundamentally sustainable. Because the ratio of those paying in and those paying out changes, the ratio of what people pay in versus what they take out will change from generation to generation (or more accurately cohort to cohort). In practice older generations will do better – but on average they also had lower incomes and longer work hours and on and on while young; most of us would not want to trade lives with them even if the pay-out they receive from social security relative to what they put in will be a little more favorable than for me and in turn I may do a little better than my children. There is no way to undo the fertility choices of our grandparents and parents, and the numbers aren’t huge in the case of the US. For more on this overall concept of “generational accounting” Laurence Kotlikoff (with various co-authors, particularly Alan Auerbach) has written widely on this topic, both for the US and for a wide range of other countries; the CBO and others have published critiques. I’ve not put in citations – you don’t want to touch this literature if you’re not a serious policy wonk or economist who needs to know, and if you are one of those, you know how to find it readily enough.
  5. In the background are much bigger numbers for healthcare, because retirees consume disproportionately more than workers, and there has been no success to date in keeping the rise in costs below the growth rate of the economy.
Mike Smitka Written in Preparation for a talk to the Lexington, VA Coffee Party on 16 August 2010