How Does QE Work? A Picture Worth a Thousand Words
The European Central Bank has started buying government bonds. Will this have further effects on interest rates?
Following the announcement of its Quantitative Easing (QE) program in January, the ECB has started buying government bonds this week. Should we expect these new purchases to have an impact on interest rates in the euro area or should we expect all of the impact to have occurred with the original announcement?
You might imagine economists have good answers to this kind of basic question but QE has always been a bit of a mystery. On the one side, you have had people who think it is an incredibly powerful tool for inflationary stimulus, e.g. the Republican-leaning economists who warned in 2010 that QE would “risk currency debasement and inflation”, warnings which have yet to be validated.
On the other side, you have many people who have trouble believing QE should affect financial markets much at all. For those trained in academic finance theory, it is somewhat troubling to think that Central Banks purchasing large amounts of an asset should change its price. Finance theory teaches that assets are priced according to their “fundamental” value so there should be a limited role for “demand” effects. As Ben Bernanke put it,
“The problem with QE is it works in practice, but it doesn’t work in theory.”
I suspect the evidence it works in practice Bernanke had in mind is a series of empirical papers by Federal Reserve researchers (e.g. this and this) but these papers still leave a lot of unanswered questions.
So, for example, should it matter at all that the ECB has started making bond purchases this week? Should the effects of the QE program have been priced in already?
Here’s a picture from a great research paper by Christoph Trebesch and Jeromin Zettelmeyer that helps to answer this question. It shows the change in yields for a set of Greek government bonds between May 7, 2010 and May 17, 2010. This is when the ECB began purchasing Greek bonds with its Securities Market Program.
The data points circled by the red line correspond to bond issues that were not purchased by the ECB, while the rest of the data points correspond to bonds that had some ECB purchases. The striking result is that only bonds that were purchased by the ECB saw falling average yields. And the more the ECB bought, the more the yields fell.
The reason this is such a great piece of research is this really isn’t how this event could have played out. You could easily imagine financial markets deciding that all Greek bonds should have had lower yields — there was a big increase in demand for these bonds and the different bond issues could be considered very close substitutes. Instead, it appears that bond purchases worked through a pure “brute force” effect — when demand for a particular issue went up, the yield for that issue was affected without spilling over into other bonds.
(Interesting side-note: The ECB did not announce at the time which bonds they were buying. We know this now because ECB later tendered these bonds for shiny new non-defaulting ones prior to the 2012 Greek debt restructuring. Not a great event for the Greek people but a nice one for academics looking for a cool finance paper.)
The paper also looks at what happens after the initial purchases: The difference in the behavior of ECB-targeted bonds and those it decided not to purchase appears to have persisted for as long as the ECB was intervening heavily in the market, up to July. So the effect described in this graph was not simply “arbitraged away” over the following weeks.
How then have the first few days of bond purchases gone? One easily available indicator is the average 10 year bond rate for AAA-rated countries as published by the ECB here. After the QE program was announced on January 22nd, this rate dropped by 19 basis points by the next day to 0.42 percent. After that it bobbed around and was 0.41 percent last Friday, March 6. After three days of ECB buying, it was 0.24 percent on Wednesday, March 11, a further decline of 17 basis points. See the graph below.
Seems like the brute force theory of QE is still doing pretty well.