Debunking the Keynesian Policy Framework: The Myth of the Magic Pendulum

Way back in early March – before we witnessed new extremes in sloppy island bailouts and malicious economist witch hunts – I promised to “wield a ruler” and clear up some fallacies about the lengths of the things. The idea is that lengths are sometimes, well, exaggerated.

I discussed two length fallacies at that time, arguing that our fiscal challenges are more daunting than they’ve ever been, while proposing a debt threshold (imagine that!) beyond which countries struggle to restore fiscal discipline without default.

In this long-delayed third pass at the Length Fallacy Challenge, I’ll consider the third question – “How long is an ‘unassisted’ economic expansion?” I’ll choose an answer from the options on the right-hand side.

length match 4a

It may seem as though a question about economic expansions doesn’t belong in the challenge, since the other questions relate to our soaring government debt and other imbalances.  The connection is that varying business cycle perspectives can have huge effects on fiscal policy decisions, as I’ll explain conceptually and then demonstrate with data.

The policy approach that no one dares to question

Consider the following argument for fiscal stimulus (and I’ll get to expansion lengths in just a moment):

In the long-term, we need to fix our public finances. We’re on an unsustainable path that needs to be corrected to protect younger and future generations. But in the short-term, we need to focus on growth. The economy stinks and people are suffering. Any attempt to lower debt in these conditions would be folly. On the contrary, the government needs to provide more stimulus to promote growth.

This is surely not the first time you’ve heard this argument. We hear at least some variation whenever the discussion gets around to our national debt.

And it’s easy to see why.

The argument allows policymakers and pundits to show concern for both long-term finances and short-term difficulties. They can appear to be far-thinking and pro-active, strategic and tactical, responsible and compassionate, and all at the same time.

In other words, the argument appeals to anyone interested in popularity, and who doesn’t want to be popular?

But what about the general public? Why do we rarely challenge it?

I’ll suggest it’s because we tend to relate the economy to a single individual, typically ourselves. When the unemployment rate is high, we think about losing our jobs and the expectation that we would interrupt long-term financial plans while unemployed. After finding work again, we would expect to reinstate those plans. And it’s easy to imagine this logical mix of long-term goals and short-term tactics applying to the broad economy as well.

Alongside basic human nature, this way of thinking helps to explain why people always look to the government to make things right when the economy’s in the doldrums.  And the public’s calls for help make it easy to justify fiscal stimulus and delay action on the debt. Policymakers can follow the simple solution of paying lip service to the need for responsibility, and then saying how foolish it would be to get started today. Their stance is seldom questioned, especially as policymakers on both sides of the aisle say the same thing. Call it the Teflon Solution.

Pendulums and magic pendulums

Unfortunately, like all things that seem too good to be true, the Teflon Solution is flawed. To see this, we have to think about the economic scenario that needs to play out for it to work. The economy needs to reach its full employment ideal and then stay there, even after fiscal policies become restrictive in the effort to reduce debt.

But do economies just recover and stay recovered, in the same way that individuals regain employment and stay employed?

The Congressional Budget Office (CBO) and Office of Management and Budget (OMB) say “yes.” When they tell us about our public debt trajectory, their projections are based on mean reversion to a perpetual state of full employment. And this assumption underpins much of the political debate about government debt. Policymakers and pundits routinely rely on CBO and OMB projections to gauge the long-term risks of budgeting decisions.

On the other hand, my answer is a big, fat “NO.”

Economies are inherently cyclical. Recoveries are helped along by the financial sector (through private credit expansion), external sector (think foreign capital) and/or public sector. And then bankers eventually overlend, businesses overinvest and/or consumers overspend, creating imbalances. When stimulus is replaced with restraint, the economy ends up back in the sickbed.

Think of it this way: When you push the economic pendulum forward with stimulus, you’re usually setting up a stronger move in the opposite direction once the stimulus runs its course. And this is the part that policymakers and economists ignore, deny or just plain lie about.

Economists, in particular, have built their most established theories on the idea that the economy isn’t a true pendulum but a Magic Pendulum. The Magic Pendulum of economic theory never swings back and forth. Instead, it always returns directly to its lowest point, where it remains still until disturbed by an external shock. At its lowest point, the labor force is fully employed and the economy is said to be in equilibrium.

Now, this naïve approach to economic theory is often criticized from both within and outside the profession, and especially after events of the past decade. The global financial crisis demonstrated once again that the pendulum tends to swing through its lowest point without stopping, and then eventually reverses direction. Stability breeds instability. But the equilibrium models that assume otherwise continue to play a dominant role in the way that economics is taught and researched. And they help to explain broad acceptance of the CBO’s and OMB’s recession-free debt projections (which, by the way, rely on assumptions crafted by economists), and the Teflon Solution, which anticipates a recovery to full employment but not the next recession.

Of course, we’d all be better off if economic theory were more accurate. If the true pendulum were indeed magic, it would make more sense to wait for a full recovery before bothering with looming threats to government finances. The Teflon Solution would work.

But in reality, the Teflon Solution helps to explain our soaring debt. It ensures that fiscal restraint occurs rarely and only for short periods of time. When the economy recovers strongly enough that restraint is put into place, it tends to swing the real life (non-magic) pendulum back into recession, or at least slow growth. And at that point, political priorities revert back to the case for stimulus.

And sometimes the stimulus only needs to wear off for the real life pendulum to reach its turning point. Remember the idiotic tax rebate that Congress served up in Spring 2008? Many Keynesian economists argued that this little gift would push the economy right back to full employment and we’d continue on our merry way. But it was obvious to more realistic folks that spending would merely jump forward for a few months and then jump right back once take home pay reverted to pre-rebate levels. Sure enough, spending collapsed in the third quarter of 2008, and once again economists’ Magic Pendulum models blew up in their faces.

Keynesianism + Magic Pendulum = Teflon Solution

Okay, the sub-header’s dorky but all economics articles have formulas, right?  I’m just pointing out that the Teflon Solution is also rooted in an interventionist or Keynesian approach to fiscal policy.  It’s the combination of Keynesianism and the Magic Pendulum that leads to a chronic imbalance between stimulus and restraint.

But I’ll leave the more fundamental problems with Keynesianism for another day, and close by throwing some data at the Magic Pendulum.  As I said above, the Magic Pendulum holds that economies remain at full employment long after stimulus is removed.  I’ll show here that they don’t.

What does the data say?

I’ll work with figures from the table below, which contains America’s business cycle history according to the National Bureau of Economic Research (NBER).

lengths 3A

The table shows that the median length of an expansion is 30 months and the average is 39 months, which happens to be shorter than the current expansion by eight months and counting.

But I’m not interested in the average of all expansions. To test the Magic Pendulum, I’d like to know about the expansions that aren’t fueled by unusual stimulus.

More precisely, I’m looking for an answer to the question stated earlier:  How long is an “unassisted” economic expansion?

I’ll screen out the expansions that were prolonged by at least one of three different types of stimulus: war, deficit spending and external debt.

War. I’ll eliminate each expansion that occurred during major wars, defined as those that lasted at least three years and cost at least 1% of GDP in the peak spending year. There were six in total, including the Civil War, World War I, World War II, Korea, Vietnam and Iraq/Afghanistan. Wartime expansions tend to be longer than peacetime expansions and account for over half of the ten expansions that lasted longer than 39 months. And the length of the wartime expansions is, of course, explained by government spending.

Deficit spending. I’ll also eliminate the expansions during which federal debt grew at an average annual rate of more than 1% of GDP, measured from the year after an economic peak to the year of the next peak. In addition to four wartime expansions that met this criteria, the mid-1930s and 1980s expansions were also helped by large increases in government borrowing.  (And the current expansion will join this group once it ends.)

External debt. My third stimulus category is based on external debt, which clearly reached a saturation point at the end of the last decade’s housing boom. The most relevant figure is net external debt, calculated by subtracting U.S. holdings of foreign debt from foreign holdings of U.S. debt. When foreigners lend us more money than we lend them, domestic spending is boosted by the additional liquidity sloshing around the U.S. To determine which expansions benefited from rising external debt, I applied the same 1% of GDP threshold that I used for the other categories of stimulus. Each of the last three expansions exceeded the threshold.

I’ve summarized the results of my screen in the table and chart below. The table matches each type of stimulus to the expansions in which it lent a hand.

lengths 3B

And the chart compares the stimulus-assisted expansions to the other expansions in the data set.  It shows that nine of the ten longest expansions can be linked to at least one of the three types of stimulus. The remaining expansions help us to understand the characteristics of an unassisted business cycle.

lengths 3C

A tale of mortality and survival

The best way to interpret the chart is to calculate mortality and survival rates for expansions that weren’t assisted by any of the three types of stimulus, in the same way that actuaries construct mortality and survival rates for people. Here are the results:

lengths 3D

Consider first that we can’t establish meaningful mortality and survival rates for the first row in the table (12 months or less), for two reasons:

  1. Periods of growth that last only a few quarters are unlikely to meet the NBER’s criteria for an expansion. For example, the 1969-70 recession included two consecutive quarters of positive growth in the middle, but the NBER chose to embed that growth within a “double-dip” recession instead of declaring a short expansion.
  2. It usually takes about a year for the NBER to declare the beginning of a recovery. In fact, the time delay from troughs to the declaration of those troughs averages exactly 12 months since 1980.

Of the remaining twenty stimulus-free expansions, only 50% survived beyond 24 months. For those that did, the survival rate (the percentage of those lasting more than 24 months that continue beyond 36 months) dropped to 30%. Two of the three remaining expansions ended in their fourth year, implying a fourth year survival rate of 33%. And the fifth year survival rate was zero out of one, or 0%.

These calculations lead to a surprising conclusion: The odds of an unassisted expansion ending in any given year were at least 50%.

To say it again differently: In 150 years of business cycles, anyone expecting unassisted expansions to continue for another year would have been proven wrong at least half the time.

Even I didn’t expect the data to be quite that convincing.

The relevance of long-forgotten history

Now, many analysts would downplay the period between the Civil War and the Great Depression, which produced many of the stimulus-free expansions.  Moreover, we shouldn’t look just to the past to predict the future.

But we also know that the long expansions of recent times were skewed by a debt binge that’s been building for decades. And unless you’ve succumbed to the happy but insidious spell of the “debt doesn’t matter” school, you understand that debt can’t outgrow the economy forever. On the contrary, it eventually becomes far more of an economic problem than an economic stimulus.

In other words, recent history may be no more relevant to the future than the largely forgotten history of the late 1800s and early 1900s.

Getting back to the Teflon Solution, business cycle history lends no support to its key premise – the idea that the economy will return to full employment and stick there, allowing ample time for debt reduction. Once stimulus is removed, expansions often struggle to continue for much longer. And if the stimulus is replaced with restraint, it seems logical that the expansion’s expected life shortens further. In other words, there is no Magic Pendulum.

What’s the typical life of an unassisted expansion?

Based on the data presented here, I’ll call it two years.

length match 4


Charles Hugh Smith discussed “The Myth of a Self-Sustaining Recovery” in a post last month on his excellent site, OfTwoMinds, and on ZeroHedge.  His self-sustaining recovery is more or less the same thing as my magic pendulum.  Here’s the original article on the magic pendulum myth.

Bookmark and Share
This entry was posted in Uncategorized. Bookmark the permalink.

One Response to Debunking the Keynesian Policy Framework: The Myth of the Magic Pendulum

  1. Mike Hihn says:

    This would be 90% shorter, and lot easier to grasp (O’m teasing), if you’d simply pointed to the obvious in NBER business cycle data.

    1) The “postwar boom at 90% tax rates” was actually … five recessions in only 16 years! The boom was invented to “prove” that WWII really, really, really worked as a giant public works project. huh-uh The economy collapsed into an immediate recession when wartime spending ended. Stimulus does indeed increase both jobs and the economy, but both the jobs and the GDO growth are temporary, plus the economy is from double-counting government spending in GDP.

    2) The 1958 recession, the most severe of the five, began two years after the Interstate Highway Act, then described as “the greatest public works project since the pyramids.” ooopsie

    3) Look VERY closely to the “peak” in March of 1933. That’s when FDR took office. The Depression was OVER, before a single penny of stimulus spending. (Then as now, it took over a year to know when a downturn has ended — to avoid double dips)

    Not to deny your great research.

Comments are closed.