Ed Dolan's Econ Blog

What Quantitative Easing Did Not Do: Three Revealing Charts

The Fed has declared an official end to quantitative easing. It is a logical time to ask, did QE work? Danielle Kurtzleben gives the honest answer in a recent post on Vox: “It’s very, very hard to know.”

Still, we do know three things that QE did not do. These are worth pointing out, especially since back when QE was just getting under way, there were people who expected that QE 2 would do all of them.

1. QE did not work according to the textbook model

One thing was never in doubt.  As the Fed added massively to its assets, QE would cause an equally massive increase in the monetary base—the sum of bank reserves and currency that accounts for the bulk of its liabilities.

Some economists used to refer to the base as high powered money. It got that name from a familiar textbook model, according to which two simple ratios link the monetary base to the rest of the economy. One is the money multiplier, which is the ratio of ordinary money (M2) to the monetary base. The other is the ratio of nominal GDP to the M2 money stock, known as the velocity of circulation of money, or just velocity, for short.

If the money multiplier and velocity were constants, then the monetary base would be high-powered indeed. Any increase in the base (which the Fed can manipulate at will) would cause a proportional increase in nominal GDP. The only thing left to determine would be how much of the change in nominal GDP would express itself as an increase in real output and how much as inflation.

The idea of treating the money multiplier and velocity as constants is not completely silly. There have been times in recent history when one or both of those ratios were, in fact, reasonably stable. For example, for the whole period from 1959 to 1992, the velocity of M2 stayed in a narrow range between 1.7 and 1.9, with ups and downs from year to year but no obvious trend. Similarly, for most of the 1990s and 2000s, the M2 money multiplier stayed in a narrow range of about 8.0 to 8.4. (Is it a coincidence that the periods of maximum stability for the ratios did not overlap? Probably not, but I’ll save that rather technical issue for another post.)

When QE came along, though, and when short-term interest rates fell effectively to zero, all hell broke loose with those deceptively stable ratios. Take a look at the following chart, which shows data for the monetary base, the M2 money stock, and nominal GDP. To make it easier to compare their trends, all three series are plotted with their value as of the first quarter of 2008 is set equal to 100.


Before mid-2008, the three series move closely together, as we would expect them to if the money multiplier and velocity were textbook-stable. Then, as soon as the Fed undertakes QE1, beginning in the second half of 2008, the series diverge. The monetary base soars as the Fed buys up vast quantities of financial assets, but the money stock barely budges. The divergence of those two series indicates a precipitous decrease in the money multiplier. Meanwhile, nominal GDP continues to fall for a year or so, and even after that, it grows more slowly than the money stock. The divergence of the money stock and nominal GDP indicates a decrease in velocity.

In short, the chart shows that in a deep slump, expansionary monetary policy is, as the saying goes, like “pushing on a string.” The Fed pushed and pushed, but the only thing it had the power to push was the monetary base, and that had very little effect on nominal GDP.

2. QE did not cause inflation

The sluggish reaction of nominal GDP to an increase in supposedly “high powered” money can be viewed as either a good or a bad thing, depending on your perspective. Back at the start of QE, not a few observers warned that QE would quickly lead to hyperinflation. That did not happen, and we can be glad it did not. However, it would have been nice if QE had boosted the growth of GDP by enough to bring inflation at least up to the Fed’s modest 2 percent target and, at the same time, to pull real output back to its potential and the job market back to full employment.

Instead, as the next chart shows, what we got was a painfully slow recovery of the real economy. The unemployment rate is still not all the way back to the 5.25-5.75 percent range that the Fed considers “full employment.” Meanwhile, inflation, as measured by the Fed’s preferred indicator, the deflator for personal consumption expenditures, peaked three years ago. Since then it has fallen well below its 2 percent target. When it looks at these data, the Fed sees enough progress to call an end to QE, but it is not as much or as fast as many of us would have liked.

3. QE was not powerful enough to overcome fiscal restraint

The policy experiments of recent years have given us a test of the relative strength of monetary and fiscal policy. Expansionary monetary policy and contractionary fiscal policy have gone at it head to head, with the outcome pretty much a draw. Fiscal policy has not been restrictive enough to derail the recovery completely, but neither has quantitative easing proved powerful enough to break decisively through the fiscal restraint.

Just how tight has fiscal policy been? The next chart shows the best overall indicator of the stance of fiscal policy, the structural primary balance, shown as a percentage of GDP. The structural primary balance is the deficit or surplus of the consolidated budgets of all levels of government as they would look at full employment, under current law and excluding interest on government debt. The current budget deficit—the indicator political discussions usually focus on—can be misleading, since it tends to increase automatically in a recession and decrease in an expansion even if there is no change in policy. The structural primary balance filters out cyclical effects to reveal underlying policy changes. A movement toward deficit shows added fiscal stimulus, while a movement toward surplus shows fiscal tightening.

If we compare this chart with the first one, we see that monetary and fiscal policies were pulling in the same direction during the downturn of the Great Recession, up to the middle of 2009. We can be thankful for that. Without QE1 and the combined fiscal stimulus programs of the late Bush and early Obama administrations, the recession would have been even deeper than it was. However, at about the same time the economy began to recover, fiscal policy shifted toward restraint, and has continued to move in that direction through this year.

What Ben Bernanke said a couple of years ago remains valid: “Monetary policy cannot achieve by itself what a broader and more balanced set of economic policies might achieve.”

The bottom line

This brief look at what QE did not do helps us understand why it is so hard to know what it did accomplish. It was an uncontrolled experiment. There was no way to apply QE to half of the economy and a placebo to the other half. Furthermore, for most of its life, the expansionary effects of QE were fighting against the contractionary effects of fiscal policy. It was a standoff, but there is no way to tell if that is because both policies were weak, or because both were equally strong. The most widely accepted conclusion about QE—that things would have been even worse without it—remains plausible, but since that is a counterfactual hypothesis that can never be conclusively tested, the debate will undoubtedly continue.

Related posts

Quantitative Easing and the Fed: A Tutorial (Slideshow)

Whatever Became of the Money Multiplier?

As We Move into Fiscal Chaos, Just How Bad is Our Fiscal Policy, Really?

20 Responses to “What Quantitative Easing Did Not Do: Three Revealing Charts”

windrivenNovember 4th, 2014 at 10:40 am

Quantitative Easing always struck me as a handout to Wall Street rather than an effort to stimulate the economy. QE never resulted in a huge upswing in lending and with high unemployment and stagnant wages, there was little demand side stimulus, especially given tight fiscal policy.

Companies don't hire on the come. They don't expand when demand is weak. Consumers don't buy when they are uncertain about their jobs or when they've been on unemployment for 30 weeks.

Economics may not have the empirical elegance and clear experimental foundation of physics. But it doesn't take a double blinded, randomized experiment to predict that QE won't do squat for the general economy unless banks are forced to use that money in ways that stimulate broadly. But then fiscal stimulus is not Wall Street's job, it is the job of fiscal policy. So why be surprised that fiddling with the water pipes doesn't make the lights come on?

toothmouthNovember 12th, 2014 at 4:56 pm

Not to mention, the two thirds of Americans that aren't fortunate to own a stock portfolio, any savings they have in a traditional interests earning FDIC saving bank has actually given them negative yields for the past 5 years, if you factor in inflation.

__Tom__November 4th, 2014 at 11:21 am

While it's true that the "money multiplier" stopped working because of the economic slump, more broadly it only works under a particular monetary regime. Such a regime employs limited excess reserves and substantial required reserves as a throttle on credit demand. In such a regime, adding excess reserves enables a many times greater increase of bank lending and broad money. That is the traditional US regime, but had been gradually weakening through loosening of reserve requirements, and was abandoned completely after Lehman. Since such a regime relies on throttling credit demand to calibrate aggregate demand, it doesn't work when credit demand is low, so its abandonment was a direct result of the slump. But such a regime is far from the global standard. Europe and many countries were already using before 2008 a regime similar to the US regime since 2008, with interest on reserves. In such regimes there is no money multiplier, slump or no. For example the Philippines, one of the fastest growing countries in the world, has an IOR regime, and thus no money multiplier.

iraleiferNovember 4th, 2014 at 2:11 pm

Ed, Seems to me that if the Fed had given the funds in spending coupons (cash) infusion to every American, for QE of 4-6% of the economy per year, one should have had growth of 4-6% plus some multiplier effect – assuming . Instead, performance has been less. Can you comment?

Secondly, the Fed could have achieved its employment goals quicker if congress had just tossed people into the "no longer looking for work" quicker – ie, cutoff benefits sooner as the jobs would not have been there and presto magic!

BenL8November 4th, 2014 at 4:30 pm

The QEs increased the excesses in financial accounts, increased the total private savings by 47% in six years to $82 trillion, while the median income continued to decline, and a financial assets bubble inflated. Laura Tyson says that concurrently fixed investments have declined. From a NYTimes article:
When measured as a share of G.D.P., this trend is even more worrisome. Net productive investment averaged roughly 4 percent of G.D.P. in the United States in the postwar period until 2000. The share plummeted after the 2001 recession and again after the 2008-9 recession, falling to a historic low of just 0.63 percent of G.D.P. in 2009.
In 2012, after four years of [so called] recovery, the share was still below 2 percent of G.D.P., less than half of its 2000 peak of 4.7 percent.

Robert Pollin correctly advocated a method to pull the string. Here's an excerpt from a Challenge Magazine article, 2010:
In the current climate, the federal government should roughly double its overall loan guarantee program—that is, inject another $300 billion in guaranteed loans into the credit market, and shift the focus of the new guarantee programs to business. Overall guarantees would therefore be about $600 billion, with a $300 billion increase from 2009. For this initiative to be effective at reducing risk and encouraging new investment, the terms will have to be generous—that is, very large guarantees, in the range of 90 percent; low or no fees on the loans; and low interest rates for borrowers.
The sticks are for the federal government to tax the excess reserves
now held by banks. This should create a strong disincentive for banks
to continue holding around $1 trillion in excess reserves." —
This article indicates to my eyes a major failure in economic theory and practice. Bernanke was too polite when he should have reminded the world that 1933 to 1937 the unemployment rate dropped from 25% to 9.6%, and it was "The Real Lesson of the Great Depression: Fiscal Policy Works", there is a Marshall Auerback essay on this at Next New Deal.

windrivenNovember 5th, 2014 at 10:00 am

Couldn't agree more. $300 billion funneled to especially small business would have spurred investment in capital equipment and hiring. Even if half of those loans ultimately defaulted the cost of the stimulus would have been (relatively speaking) inconsequential.

Alexandre HaninNovember 7th, 2014 at 5:49 pm

MMTers (and others) predicted from the start that QE would not be inflationary, and explained why.

Banks are not reserved-constrained. When a credit-worthy borrower wants a loan, the bank gives him a loan without even checking whether it has enough reserves. If the bank realizes afterwards that it doesn’t comply with reserve requirements, it can find reserves on the interbank market (I think it has 15 days to do so, but I could be mistaken).

If too many banks are making too many loans, then the short-term interest rate starts rising. However, as the central bank targets a specific interest rate, it has to inject more liquidity into the system immediately to hit its target. So the CB has no choice but follow the market.

The CB can control either the short-term interest rate (that’s how it works today and how it works usually) or the monetary base (what Volcker tried, without much success, in the 80s), not both.

If banks can always find the needed reserves, then QEs cannot allow them to lend more. QEs simply lower the short-term interest rate. And the fear that all those reserves in the banking system will suddenly flow out and cause hyperinflation once the economy starts growing in earnest are misplaced.

__Tom__November 9th, 2014 at 4:38 pm

I'm surprised Ed agrees here, because this MMT dogma is wrong and completely misses the point of why QE didn't cause inflation.

What Hanin is describing is a monetary regime like the US had up to 2008, where the supply of excess reserves and reserve requirements are used to throttle credit demand and thus directly target the overnight interbank rate and indirectly target inflation and employment.

In such a regime, the central bank is indeed committed to supplying as much reserves as banks want to borrow at the targeted overnight interbank rate. However the rate is nonetheless an effective throttle as it applies to credit demand from ultimate borrowers. A 1% target is a nearly open throttle. A 6% target is a tighter throttle. Yes, with a 6% target healthy banks can still always find reserves, but at the higher rate of about 6%, and thus they will charge higher rates on their loans, and thus there will be less demand for credit from borrowers. That's how the central bank keeps its reins on credit expansion and inflation in such a regime.

According to MMT dogma, QE doesn't cause inflation because the supply of excess reserves don't matter to the rate of credit expansion. That's absolutely not true. The supply of excess reserves determines prevailing overnight interest rates, which mainly determines prevailing bank lending rates, which is hugely influential on the pace of credit expansion and inflation.

If QE had been applied in say 1997, the result would have of course been more inflation. Short rates would have dropped to near zero, credit expansion would have boomed.

The reason QE didn't cause inflation in 2009 on is that credit demand was too low to cause inflation even with excess reserves completely unthrottled.

Ed Dolan EdDolanNovember 9th, 2014 at 5:24 pm

That's not my understanding of the MMT "dogma" as you call it, or of what Hanin was getting at.

My understanding is that MMT in general and Hanin in particular are saying is that if the central bank pursues and interest rate target, then lending is not supply-constrained at that rate. (The idea that it was supply constrained (reserve constrained) was the essence of the "textbook model" as I call it). I don't think either you or I disagree there.

Lending is always demand constrained in the sense that it takes a willing borrower as well as a willing lender to enter into a loan contract. I have seem some MMTers who appear to say that borrowers pay no attention to interest rates when they decide whether to borrow or not, but I don't think that was Hanin's point. It is an area of MMT that I have more trouble with than the supply constraint part, which I think was his point.

__Tom__November 9th, 2014 at 9:01 pm

Actually the textbook model is correct for the pre-2008 monetary regime. Supply of reserves was in fact constrained, to the extent consistent with the central bank's rate target. Constraining reserves is the mechanism by which the central bank raises rates. For example if the central bank wanted to tighten from 5% to 5.25%, the way it would do that is to constrain the growth of reserves to less than they would grow if the central bank wanted to maintain a 5% rate. When reserves supply is truly unconstrained, and there's no IOR, overnight interbank rates drop to near zero.

I think it's fair to call MMT a dogma and I think it's fair to assume Hanin was invoking and following that dogma given his intro sentence and the text of his comment. The false claim that reserves supply was unconstrained in the pre-2008 US monetary regime is at the center of that dogma. I understand you want to be as generous as possible, but I think it's important especially for educators to resist false claims about how the monetary system works from wherever they come. Cheers.

Ed Dolan EdDolanNovember 10th, 2014 at 7:07 am

Supply of reserves was in fact constrained, to the extent consistent with the central bank's rate target." I think that is just the point that MMT makes. In fact, it is wrong to link this point to MMT, it is pretty much universally accepted among monetary economists. The Fed can set a target for the interest rate, or reserves, but not both. Even pre-2008, it set the target for interest rates. Until the interest rate target was changed, reserves were not separately constrained or targeted. As far as I know, the Volcker experiment was the only attempt to separately target reserves, and it did so at the expense of letting the interest rate go wherever it would.

As for MMT, yes, it is a dogma for some (by a dogma, I mean a set of phrases learned by rote without understanding or use of critical facilities), just as praxeology, logical positivism, Marxism, neoclassical micro, etc. can become dogmas. For others they are meaningful theories. I have no idea what Hanin thinks, since I don't know him. Maybe he will speak for himself if he is still following this thread.

As for the textbook model, if you have read any older editions of textbooks lately, you will see that it comes in different flavors, but they all contain one or more of the following four flaws: (1) they present a "monetary expansion" process in which not just the banking system, but individual banks are reserve constrained; (2) based on that, they present a money multiplier that is treated as a constant at least for the purpose of problems that are assigned to students, even if they make verbal caveats in the text; (3) ditto for velocity; (4) they acknowledge the use of an interest rate as an operating target but they represent it as a mere mechanical tool for hitting a quantity of money target. Real monetary economists, whether MMT or not, have realized the limitations of those concepts for years, but the old textbook model is by no means dead in the classroom.

__Tom__November 11th, 2014 at 10:31 am

Okay, we can just agree to disagree on what MMT teaches. Everything I've seen from them is that reserves are not constrained at all ever in an interest-targeting regime. That is, I've never seen any MMT proponent ever mention or concede that the interest target is a kind of relative constraint on the money supply, in the sense that I just explained (eg a 5.25% rate would result in less money supply than a 5% rate, all else equal).

On the textbook models, you're basically right in your characterization, though 1) and 2) are more oversimplifications than important flaws. The fact is broad money supply and base money supply did behave fairly close to that model for decades. The model didn't explain why the m2/m0 multiplier actually ranged from around 7 to 12, but it gave a fairly clear representation of what was roughly going on.

I've never encountered anyone who took the model so literally that they assumed that the multiplier really was constant, or that banks really had to check their reserve balances or top them up to sufficient levels before they made loans. Kicking up a big fuss over those simplifications in the textbook model always seemed to me to be somewhat facetious.

Where I find that very many people are confused by the old model is that they still believe that it applies almost universally, with the exception of during "liquidity traps." This seems more a flaw of how the model was taught than of the model itself, but I constantly run into people who don't understand that the model only describes one particular monetary policy regime.

As for your 3) and 4), I see those as larger flaws of mainstream 20th-21st century economic theory, incorporated in various models, including also IS-LM . The assumption that money supply governs spending, or put another way of a meaningful, mean-reverting V, runs throughout the neo-Keynesian synthesis. Although you could argue that V was similarly range-bound for decades, the modeling of V is explicitly meant to be universal and not tied to any particular monetary regime. Thus in my view it's much more than an oversimplification; it's extrapolation of an untrue alleged universal from limited observation.

That problem hasn't really been addressed in mainstream theory. Even Krugman while leading the attack on base-money-growth-equals-inflation does so by presenting the "liquidity trap" as an extreme case of the old IS-LM model, which implies that money supply and V will return to playing important roles as soon as we exit from the trap.