Download this issue brief (pdf)
Any attempt to address long-term federal budget deficits will require answering some big, fundamental questions. How much spending is enough? How much revenue is too much? Usually, the answers to these questions are couched in terms of a share of the wider economy. In other words, how much of our combined national income do we want the government to spend? And therefore, assuming we want a balanced budget, how much of our income are we willing to pay in taxes?
Given the dramatic implications of setting these overall targets, it is not surprising that many people turn to history as a guide—especially when faced with a problem of the magnitude and complexity of the current deficit dilemma. Indeed, it is common to see lawmakers, pundits, and commentators cite the last 60 years as a factor in their thinking about what to do between now and 2070.
Case in point: Erskine Bowles, the co-chair of the President’s Commission on Fiscal Responsibility and Reform, seems to have history in mind when he recently proposed 21 percent of gross domestic product—the total value of all goods and services produced in our economy—as an appropriate target for both spending and revenue. Twenty-one percent of GDP is close to the historical average of federal spending.
The most common suggestion, though, especially from conservatives, is a target of 18 percent of GDP, which is the approximate average amount of revenue collected by the federal government since 1950. That certainly sounds like a compelling statistic. It implies that the amount of revenue the federal government raises has been constant and sufficient for the past 60 years. If 18 percent of GDP has been good enough for us since World War II, why change it now?
But there are many good reasons to be skeptical that averages from the last 60 years are appropriate benchmarks for the next 60. First and foremost, these averages are highly misleading. First, an average does not tell the whole story but rather obscures trends. In fact, that “18 percent” obscures a clear rise in tax revenues, decade by decade, as the federal government took on more responsibility for the health care of senior citizens and as health care costs rose more generally. Second, 18 percent of GDP has actually been an insufficient level of revenue to pay for expenditures for some time now. Locking it in for the future would mean locking in deficits.
More importantly, it’s simply wrong to try and budget for the future by looking backwards and trying to shoehorn future needs into whatever the past levels have been. Instead, we should be trying to determine broadly how much public investment will be required as we move deeper into the 21st century, and then how do we pay for those investments in the most efficient way possible. Why should we even consider the average from the past 60 years as an appropriate constraint until 2070? Certainly everyone agrees that times and circumstances have changed, and that the federal government should, presumably, change with them.
Tackling our looming budget challenge is going to require tough decisions, and we should welcome guidance from any quarter, including our recent history. But nothing in that history suggests that setting an arbitrary spending and revenue target is the right approach to achieving a balanced budget. Using a flawed metric of a 60-year average as that arbitrary target would merely compound the mistake, not least because historical averages are not even particularly good descriptions of our past, let alone good guides for our future.
The average is one thing, the trend is another
The federal government has, over the past 60 years, collected an average of about 18 percent of GDP worth of annual revenue, and that total federal expenditures have averaged about 20 percent. But what do those averages really mean? Does that mean that revenues and spending as a share of the economy have been stable and constant over time? Actually, no. These averages mask trends within the data, and they also convey a false impression of stability.
If, instead, we look at average tax revenue on a decade-by-decade basis, a very different picture emerges. Rather than collecting a stable, flat 18 percent of GDP, this view reveals that revenue, as a share of GDP, increased steadily from the 1950s until the end of the century. More specifically, from 1950 to 1959, the federal government collected an average of 17.2 percent of GDP in revenue. In the 1960s the average was up to 17.9 percent of GDP, and then increased again, albeit slightly, in the 1970s. The annual average then jumped by about a quarter of a percentage point in each of the following two decades, rising to 18.5 percent of GDP in the 1990s. Overall, then, from the 1950s through the 1990s, the average revenue collected, as a share of GDP, increased by more than 1.3 percentage points.
Revenue rose because, throughout most of this period, expenditures were rising too. Just as with revenues, spending rose, as a share of GDP, in each decade from the 1950s through the 1980s. Much of this rise started in the late 1960s with the creation and then expansion of Medicare and Medicaid as the federal government took on greater responsibility for the medical care and retirement needs of senior citizens. This new role led to an increase in spending. In fact, in the 16 years between 1950 and the creation of Medicare and Medicaid in 1966, federal expenditures averaged just 17.8 percent of GDP, far less than the overall, 60-year average. After the creation of Medicare and Medicaid, spending, as a share of GDP, rose to 20 percent of GDP in the 1970s and 22.2 percent of GDP in the 1980s.
Conversely, in the 1990s government spending as a share of GDP fell to an average of 20.7 percent. This decline was the result of a combination of the Clinton administration’s commitment to balancing the budget, the opportunity to reduce defense spending at the close of the Cold War, and extremely robust economic growth. The 1990s reversed a 40 year spending trend.
Importantly, though, President Clinton continued the revenue trend while reversing the spending trend by raising taxes in his 1993 budget plan. As a result, revenue grew from 17.5 percent of GDP in 1993 to 20.6 percent in 2000. The consequence of these policies of reduced spending and higher revenue was, of course, a historic four-year period of budget surpluses.
President Clinton’s successor, however, chose a different path. Where President Clinton continued the 50-year trend of increasing revenues, President Bush’s signature economic policy was to repeatedly cut taxes. Had the 50-year revenue trend continued, the federal government would have collected an average of around 18.7 percent of GDP through the first decade of the new century. Instead revenue dropped dramatically. Rather than 18.7 percent of GDP, the federal government actually collected an annual average of just 17.6 percent from 2000 to 2009.
It’s also worth mentioning that President Obama’s budget plan includes something of a return to the pre-Bush historical revenue trend. Had the 50-year pattern of increased revenues in each decade continued throughout the 2000’s and into this decade, then we would expect federal revenues to average about 19 percent of GDP over the next 10 years. Revenue in President Obama’s budget averages 18.3 percent from 2010 through 2019, and just below 19 percent from 2012 onward (after excluding the unusually low-revenue years of the Great Recession and its immediate aftermath).
The suggestion, therefore, that federal revenues and spending have remained constant at about 18 percent and 20 percent of GDP, respectively, for the last 60 years is simply false. Beyond the year-to-year fluctuations, there is a clear pattern of higher revenue in each new decade. The pattern held for five decades, and was only broken by the massive tax cuts implemented by President Bush. President Obama’s proposals would begin to bring federal tax revenues back into line with the historical trend, after a decade of abnormally low revenues.
Raising “average” revenue resulted in deficits
Besides being badly misleading, focusing only on the average amount of tax revenue collected also obscures the fundamental fact that during that same 60-year period, the federal government routinely ran budget deficits. In fact, in 51 out of the past 60 years, and 45 out of the past 50 years, the federal budget was in the red. However much revenue the government was collecting during many of those years clearly wasn’t enough.
Furthermore, there were only 9 years over the past 60, in which 18 percent of GDP in revenues would have produced a balanced budget. And all 9 of those years occurred before 1967. In other words, if the federal government had hewed exactly to the post-war average in each of the last 60 years, then we would currently be experiencing our 44th straight year of budget deficits.
The only reason we are not currently in our 44th straight year of deficits is because there were years in which revenues were significantly higher than the average—high enough to produce balanced budgets. In the five years of balanced or surplus budgets since 1960, revenues averaged 19.9 percent of GDP. In the four most recent years of budget surpluses, 1998 to 2001, revenues averaged 20 percent of GDP.
Simply put, 18 percent of GDP has been an insufficient level of revenue for some time now. To accept 18 percent of GDP as the “normal” amount of revenue for the federal government to raise is to accept the attendant deficits. Of course, if we want to actually balance the budget, history shows us that higher revenues are necessary.
The last time 18 percent was sufficient, the world looked a bit different
1966 was the last year in which revenues totaling 18 percent of GDP would have produced a balanced budget. And the world, the country, and the federal budget were all quite different then.
From a budgetary perspective, the biggest difference between the world before 1966 and the world after is the existence of Medicare and Medicaid. In July of 1965, Congress created these two health programs, and it wasn’t until 1967 that the federal government began expending any significant resources on them. In 1966, total federal spending on health care came in at less than $17.5 billion, in 2009 dollars, or only 0.3 percent of GDP. Just five years later, with Medicare and Medicaid implementation well under way, health spending passed $70 billion, or 1.2 percent of GDP.
Since then, resources for Medicare and Medicaid have continued to grow as the population ages and as health care costs rise. Last year, total federal spending on health care surpassed $760 billion, or 5.3 percent of GDP. The recently passed Affordable Care Act is a good first step toward bringing those costs down, but if it is not implemented aggressively, those trends will continue as well.
Health care is not the only area in which the federal commitment has changed dramatically since 1966. That year, Social Security cost less than $140 billion (calculated in 2009 dollars), or 2.7 percent of GDP. This past year, Social Security expenditures totaled more than $680 billion, or 4.9 percent of GDP. Similarly, in 2009 we spent roughly ten times as much on federal law enforcement as we did in 1966, after taking inflation into account. In 1966, the government spent only $9 billion for veterans’ hospitals and medical care, compared to $42 billion last year.
Some other government programs have actually experienced substantial declines. We’re now spending far less, for example, on the space program than we were 45 years ago, both in real terms and as a share of GDP. We’re also spending somewhat less on international development and humanitarian assistance.
And then there are dozens of other programs and services that didn’t even exist in 1966. These include the Transportation Security Administration, Head Start, Supplemental Security Income, the Special Supplemental Nutrition Program for Women, Infants and Children, the Children’s Health Insurance Program, the School Breakfast Program, and the Corporation for National and Community Service, just to name a few. These new programs arose in response to new needs, such as the need for greater airport safety or as a better way to address enduring problems such as childhood poverty.
It should not be shocking that the federal budget now is vastly altered from what it was nearly five decades ago. It should be different. The challenges we face are different, our national needs are different. Even the demographics of the American people are different. Take the age distribution of Americans as one important example. In 1960, the median age was under 30 years old. Now it’s closer 37. In 1960, only 5.7 percent of the population was over the age of 70. Now it’s 9 percent. Is it any wonder, then, that expenses relating to the care and retirement needs of senior citizens have gone up?
Looking forward, this trend is only going to continue. The Social Security trustees, for example, expect that the percentage of the population over the age of 65 will rise from 12.9 percent this year to 15.9 percent by 2020, and to 19.3 percent by 2030. And the costs of caring for senior citizens have risen substantially over the past decades, as have health care costs generally. An older population means there will be more pressure on the federal budget than there was in 1950. It’s a demographic certainty.
The contours of the federal budget the last time 18 percent of revenue would have actually resulted in balance are almost unrecognizable compared to those we have now. This is as it should be. A lot can change in four and a half decades. What is odd, however, is how many people seem to think it’s a good idea to use the now-distant past as a constraint on our future budgeting decisions.
2054 will definitely not be the same as 1966
If 18 percent of GDP in tax revenue hasn’t been enough for the last 44 years, what makes us think it will be enough for the next? Just focusing for the moment on the next ten years, what would it really mean to try and balance the budget in the coming decade with just 18 percent of GDP in federal revenue? It would mean massive, enormous, draconian cuts.
Using the president’s most recent budget plan as a baseline, by 2020 federal spending is expected to be just over 25 percent of GDP. If we intended to balance the budget by then, using only 18 percent of GDP in revenue, federal spending would have to come down by 7 percentage points of GDP, or approximately $1.6 trillion. That’s equivalent to a 28 percent across-the-board cut.
But we can’t actually cut all federal spending by 28 percent. We certainly can’t stop paying interest payments on the debt. But if we tried to cut $1.6 trillion in spending from absolutely everything else, then we’d need a 34 percent reduction. And “absolutely everything else” means just that. We’d have to cut 34 percent from, for example, veteran’s benefits, highway construction and maintenance, law enforcement efforts, homeland security, education funding, and science and health research. It would also mean cutting Social Security, Medicare, and Medicaid all by more than a third.
But if we further believe that it is both unlikely and wrong to ask Social Security beneficiaries to accept just two-thirds of their earned benefits, and we therefore exempt them from the 34 percent cut, then the rest of budget would have to be slashed by 45 percent. And are we really likely to cut Medicare almost in half? If the answer is no, then the cuts to everything else would have to exceed 60 percent. What about defense? What about veterans? What about school lunches? Are we going to cut all of these by 60 percent?
Recently, lawmakers have taken to calling for cuts or freezes to a very specific slice of the federal budget: nonsecurity discretionary spending. This category deliberately excludes Social Security, Medicare and Medicaid, veterans’ benefits, homeland security, and defense. It does so precisely because these things are difficult and unpopular to cut.
But in 2020, nonsecurity discretionary spending will be less than $500 billion. In other words, with only 18 percent of GDP in revenue, we could eliminate absolutely everything in this category and we’d still have a $1 trillion deficit. Just to be clear, that would mean no more federal support for education, no more highway construction at all (and no more maintenance of existing highways), no more National Institutes of Health or National Science Foundation, no more Smithsonian, no more national parks, no more Food and Drug Administration, no more job training programs, no more Federal Aviation Administration, and no more Consumer Product Safety Commission.
In fact, we could even go one step further. We could eliminate all discretionary spending, not just “nonsecurity” discretionary spending. That would axe the FBI, the Coast Guard, veterans’ health care, and of course, the entire Pentagon. But even this wouldn’t be enough. If revenues equal just 18 percent of GDP in 2020, getting rid of every single dollar of discretionary spending would still not balance the budget.
Clearly, 18 percent simply isn’t going to cut it. How about 20 percent, the historical average of spending? Obviously, the cuts would be smaller, but cutting 5 percent of GDP still means finding $1.1 trillion in reductions. Again, eliminating all nonsecurity discretionary spending would not be enough. We’d still need to cut over 60 percent from security spending, including the Pentagon. Alternately, $1.1 trillion could be found by slashing Medicare, Medicaid, and Social Security by about 40 percent.
Even if we were willing to live with these were cuts, our current tax code—the one that congressional conservatives have pledged to maintain—does not produce 20 percent of GDP in revenue. In fact, it doesn’t even produce 18 percent of GDP. In 2000, before the first of the many Bush tax cuts, federal revenue equaled 20.6 percent of GDP. Over the next ten years, Congress has passed a veritable smorgasbord of tax cuts, with most of them skewed heavily to the wealthy. As a result, from 2001 to this year, average annual revenue amounted to just 17 percent of GDP, a full point below the vaunted “historical average.”
Using Congressional Budget Office projections, we can estimate what our current tax code would raise going forward. If all expiring tax provisions are maintained, including the Bush tax cuts for the richest 2 percent of Americans, then annual revenue will average just 16.7 percent of GDP over the next 10 years. Thus, even if we ignore all of the major drawbacks to using 18 percent or 20 percent of GDP as a constraint on federal revenues, the fact is that we still need tax increases just to reach those levels.
Misleading, arbitrary, and insufficient
The federal budget deficits that we face going forward present a considerable challenge. They are the product of an aging population, rising health care costs that will take at least a decade to bring under control through the Affordable Care Act, the damaging fiscal legacy of the Bush administration, and the effects of the recent massive recession. Getting the budget onto a more sustainable path is going to be a near-Herculean task.
Making it even more difficult are those who suggest we should hew to some arbitrary level of revenue pegged to an historical average that, upon review, is hopelessly flawed. Yes, 18 percent is the average amount of revenue, expressed as a share of GDP, raised by the federal government over the past 60 years. But that number offers no utility as a guide to our future fiscal choices between now and 2070. It is misleading, it is arbitrary, and it is woefully insufficient. Twenty percent, the average amount of spending as a share of GDP, is little better.
Trying to squeeze the demographic, health care, and other challenges of the future into the constraints of the past will, invariably, result in poor outcomes. To see the folly of this kind of thinking, just imagine what post-war America would have looked like if we had forced ourselves to maintain pre-war levels of spending and revenue. From 1930 to 1940, outlays averaged just 8.1 percent of GDP. Revenue was just 5.2 percent of GDP. Could we have won the Cold War with those levels? Put a man on the moon? Cut poverty among our elderly by two-thirds? Grow the economy more than five times over?
We are going to have to make some hard choices going forward, and it is understandable, given how hard those choices are, to look for some kind of guide. But confining ourselves to an average from 1950 through today means limiting ourselves to the past. Fundamentally, we need to make budget decisions based on our current and future circumstances, not our past ones.
Michael Linden is the Associate Director for Tax and Budget Policy at the Center for American Progress.
Download this issue brief (pdf)