Quantitative Easing

How quantitative easing works

In normal times, a central bank stimulates the economy by cutting its short-term policy rate — the federal funds rate in the US, the deposit facility rate in the eurozone, the call rate in Japan. Once that rate is at zero, the conventional lever is exhausted: you cannot cut a 0% rate to encourage more borrowing.

QE works on a different point in the yield curve. The central bank announces it will buy hundreds of billions of dollars of long-term bonds — Treasuries with 5- to 30-year maturities, mortgage-backed securities, sometimes corporate bonds. The mechanics:

1. The central bank credits a primary dealer's reserve account at the Fed with new reserves — created with a keystroke.
2. The dealer transfers the bonds to the central bank.
3. The central bank's balance sheet now holds more long bonds; the dealer holds more reserves.
4. The supply of long bonds available to the private market has shrunk; their prices rise; their yields fall.
5. Lower long yields drag down mortgage rates, corporate bond yields, and any other rate priced off long Treasuries.
6. Cheaper long-term financing encourages investment, refinancing, and durable spending.

Three transmission channels operate in parallel:

• Portfolio rebalancing. Investors who sold the bonds now hold reserves; they rebalance into other assets, lifting equity, real estate, and corporate bond prices.
• Signaling. A credible QE program tells markets the central bank is committed to keeping rates low — even after QE ends, the policy rate will probably stay near zero longer.
• Duration absorption. The central bank takes long-duration interest-rate risk onto its own balance sheet, freeing private investors to take other risks.

The Fed QE timeline, 2008–2022

Round	Period	Assets bought	Approximate size	Balance sheet end
Pre-crisis	—	Treasury bills	—	≈ $0.9T
QE1	Nov 2008 – Mar 2010	Agency MBS, agency debt, Treasuries	≈ $1.75T	≈ $2.3T
QE2	Nov 2010 – Jun 2011	Long-term Treasuries	$600B	≈ $2.9T
Operation Twist	Sep 2011 – Dec 2012	Sold short, bought long Treasuries	$667B reshuffled	≈ $2.9T
QE3	Sep 2012 – Oct 2014	MBS + Treasuries, open-ended	$85B/month → $1.6T total	≈ $4.5T
QT (taper)	2017 – 2019	Allowed runoff	—$700B	≈ $3.8T
COVID QE	Mar 2020 – Mar 2022	Treasuries + MBS, $120B/month	+$4.6T	≈ $8.9T
QT (Bernanke-style)	Jun 2022 – present	$95B/month runoff	—$2T as of 2026	≈ $7T

Other major programs ran in parallel: the ECB's Asset Purchase Programme (€2.6T, 2015–2018, restarted as PEPP for COVID), the BOJ's Quantitative and Qualitative Easing (QQE, ¥80T/year from 2013, with yield-curve control from 2016), and the Bank of England's gilt purchases (£895B at peak).

Conventional rate cuts vs QE

	Conventional rate cuts	Quantitative easing
Tool	Set the overnight policy rate	Buy long-term assets
Yield curve point	Short end	Long end
Balance-sheet effect	Minimal	Large; can multiply 5–10×
When usable	Anytime above ZLB	Especially at the ZLB
Transmission speed	Fast (days to weeks)	Medium (weeks to months)
Reversibility	Symmetric — just raise rates	Asymmetric — selling bonds is harder than buying
Side effects	Predictable, well-understood	Asset-price distortion, inequality concerns
Political controversy	Low	High

Counterarguments and the inflation that didn't happen

The most cited prediction about QE was that it would unleash hyperinflation. In November 2010, 23 economists, fund managers, and political figures — including Niall Ferguson, John Taylor, and Cliff Asness — published an open letter in The Wall Street Journal warning that QE2 risked "currency debasement and inflation." Peter Schiff and Jim Rogers made similar predictions. None of it happened.

Three reasons the inflation forecast failed:

• The multiplier collapsed. Banks parked the new reserves at the Fed instead of lending them out. The M1 multiplier fell from about 1.6 in 2007 to under 0.8 by 2014, almost exactly absorbing the base-money increase.
• Velocity dropped. M2 V went from 2.0 in 2007 to 1.4 in 2015. The MV side of the equation MV = PQ stayed roughly flat even as M ballooned.
• IOER changed bank incentives. The Fed began paying interest on reserves in October 2008. With a positive risk-free yield on reserves, the marginal incentive to lend instead of hold dropped sharply.

The 2022 inflation breakout is sometimes used to argue QE finally worked as feared. The timing is wrong: the COVID inflation came two years after the COVID QE, after fiscal transfers landed directly in household checking accounts (a much more inflationary channel than QE), and during a global commodity-and-supply-chain shock. QE was a contributing factor; it was not the sole cause and probably not the dominant one.

The opposite critique — that QE was insufficient — comes from Paul Krugman, Brad DeLong, and the post-2008 Keynesian school. Their argument: QE is a poor substitute for lower rates because it operates through indirect, leaky channels. The Fed should have done more, sooner, and in larger amounts, and combined it with explicit fiscal coordination.

A third critique is that QE worsens inequality. By boosting asset prices — equities, real estate, bonds — QE benefits the asset-rich more than the wage-dependent. Empirical studies (BIS Working Paper 599, 2017; Bunn et al., 2018) find this effect is real but modest in magnitude, with the income-side benefits of avoided unemployment outweighing the wealth-distribution effects for lower-income households. The political optics, however, have been damaging.

Variants and adjacent tools

• Helicopter money. Friedman's 1969 thought experiment: print money and drop it on households directly. Bypasses the banking system and the multiplier collapse, but has no exit and crosses the central-bank–fiscal boundary.
• Yield-curve control (YCC). The BOJ's 2016 innovation. Instead of buying a fixed quantity of bonds, the central bank pins a target yield (e.g., 0% on the 10-year JGB) and buys whatever quantity is needed to defend it. Can be much smaller in volume than open-ended QE.
• QE with corporate bonds. The ECB included corporate bonds in its CSPP (2016); the Fed crossed this line during COVID with the SMCCF. Risk: blurs the line between monetary policy and credit allocation.
• Operation Twist. Sells short-term bonds and buys long-term ones, lengthening the average maturity of central-bank holdings without expanding the balance sheet. Used by the Fed in 2011–2012.
• Quantitative tightening (QT). The reverse of QE: let the balance sheet shrink as bonds mature without reinvestment, or actively sell holdings. The Fed did $700B of QT in 2017–2019 and started a faster Bernanke-style runoff in mid-2022.
• MMT critique. Modern Monetary Theory argues QE is largely cosmetic — an asset swap that does little for nominal demand. The real lever, in MMT, is direct fiscal spending; QE merely manages bond-market plumbing.

Common pitfalls when reasoning about QE

• Calling QE "money printing" without qualification. QE creates reserves, not currency in circulation. Whether reserves become broad money depends on bank lending behavior — which post-2008 was weak.
• Forecasting inflation directly from balance-sheet size. The relevant variable for inflation is broad money × velocity, not the central-bank balance sheet. The 2008–2019 period proved a 5× balance-sheet expansion can coexist with persistent inflation undershoot.
• Ignoring the exit problem. A bond bought during QE has a finite maturity and a market price. Selling it back when conditions change can move yields sharply — the 2013 Taper Tantrum was a forecasting accident.
• Treating QE as costless. The central bank earns interest on its bond holdings but pays interest on reserves it issued. When IORB rises above bond yields (as in 2022–2024), QE programs run substantial accounting losses, eventually transmitting to lower remittances to the Treasury.
• Generalizing across economies. QE worked differently in the US (deep, liquid, dollar-reserve currency), the eurozone (fragmented sovereign debt, no unified treasury), and Japan (chronic ZLB, structural deflation). Lessons from one do not transfer cleanly.
• Confusing QE with rate cuts. QE works mainly through the long end of the curve and asset-price channels. Rate cuts work mainly through the short end and credit channels. They complement but do not substitute for each other.