New "Economic Synopsis" from the St. Louis Fed: "Has QE Been Effective?", by Daniel L. Thornton. The bottom line: "The analysis presented here suggests that QE had little or no effect in reducing long-term yields relative to what they would have otherwise been... If QE did not significantly reduce long-term yields relative to what they would have otherwise been, it cannot have increased output or employment either."
The study referenced looked at the relationship between Treasury yields and other sovereign yields, and it found that there was no unusual movement courtesy of Quantitative Easing (QE).
Other studies by other Fed economists found that QE was effective, based on "announcement effect" event studies. What happened was that yields fell on the days that QE measures were announced. The highly sophisticated methodology was to add up all those changes and claim that QE reduced rates by that total, which was relatively large. (The fact that this was the answer that the top brass at Fed wanted to hear at the time may or may not have been a coincidence.)
However, any serious thought about the topic told you the "announcement effect" methodology was full of holes. Even if the markets had a knee-jerk reaction on those days, the markets would return to discounting the expected path of short rates. The "announcement effect" shocks would decay as rates drifted back to where the expected path was thought to be.
To explain further, I used to do event studies for the "June 1"/"December 1" effects in the Canadian bond market. On the first day of those months, the duration of the Canadian index would jump as the result of bond coupon payments and bonds dropping out of the index (since those dates were the standard maturity dates of long-dated Government of Canada bonds). What you saw is that since people had to buy duration, there was typically a dip in yields on those days.
So in other words, every time the first of June and the first of December rolled around, bond yields would typically drop. If you used the same methodology that "proved" QE worked, Government of Canada yields would be expected to be negative, as they keep getting knocked down by this effect every six months.
Needless to say, that has not happened (yet). What happens is that the dip in yields on the first is reversed on dates on either side of the event date. (Of course, the markets are going up and down as the result of other factors, so it's hard to give exact numbers, even if you have good quality Canadian market data.)
Other bond markets have month-end duration extension effects. (Bond indices outside Canada generally juggle their constituents only at month end.) I believe that you end up with similar games being played around the month-end dates with big duration extensions. (I never looked at the effect in the other markets that closely).
The only way to properly measure the effect of QE is to get your hands on a reliable, very accurate bond valuation model. You could then see whether bond yields are deviating from this model when QE was brought in. I believe that you will probably be able to find this model in a crate next to the one in which the U.S. Government stored the Ark of the Covenant. Otherwise, all you can do is compare yields to where you think rate expectations should be. For myself, I see no evidence that yields are markedly out of line with rate expectations.
(h/t Bruce Bartlett - @BruceBartlett)
(c) Brian Romanchuk 2014