The facts.  After hitting 0% nominal interest rates in the short term Federal Funds market, the Fed purchased almost $3 trillion longer-term assets (mortgage backed securities guaranteed by Fannie Mae and Freddie Mac, and longer-term Treasury bonds guaranteed by the Federal government) with newly created reserves.  The money supply (M1) did not increase with this huge program, nor did inflation.

The problems with the QE program, in order of seriousness and relevance:

• The program might eventually generate more inflation than would have happened.
• The Fed may be less able to implement a similar QE program in a future financial crisis.
• If the Fed decides to unwind the program, and sell the assets, that might produce financial instability.
• There is some probability that the assets will not yield the promised coupon payments and principal (i.e. MBS might default, and Fannie Mae might be insolvent and so not guarantee; or the U.S. Federal government might default, a la Greece).  The Fed itself would then not be solvent.
• The program may have contributed significantly to growing inequality “the greatest backdoor Wall Street bailout of all time.”
• There may be some appearance or reality of corruption, where assets of some sellers are treated preferentially and assets of other sellers are not purchased, in return for some implicit or explicit quid pro quo (revolving door, etc.)

Given the possible problems (costs) of the program, it is important to gauge the benefits of the program.  The rationales for the program were two-fold:

• Improve stability in the financial sector
• Lower long-term interest rate thus stimulating greater investment and consumer spending

The main alternatives to the program?

• Commit to zero short term interest rates for long time periods (i.e. five years) with gradual raises after that
• Shift existing assets to long term without adding assets
• Commit to a higher rate of inflation (say 4%)

So what were the effects of the program?  It has to be said that it is pretty much impossible to tell, since we have a sample of one country over one time period (the United States over the period 2006-2015) and the country was in a financial crisis and subsequent recession that was the whole reason for the program.  Hence there is no credible counterfactual of what would have happened to financial stability, interest rates and output/employment in the absence of the program. That is, there is no “control” sample to tell us what would have happened to a country like the U.S. facing a similar crisis that did not implement QE.  Pescatori and Turinen highlight another problem with research on the effects of QE, which is that long-term real interest rates appear to have been declining over time, and appear likely to continue to decline over time, so the counterfactual is even harder to know: the counterfactual is not a “return to normal” time period, rather it is a “never seen before” decade of negative real interest rates.

So studies of the effectiveness of the program have to focus on short-term effects (what if we purchase more assets this week compared with last week), or effects around the timing of major events in the program (what happens when program is announced or launched), or comparing with Eurozone (which is not that comparable to U.S. but is the only control possible), or effects generated by a simulation of an economy (of which there are dozens, each similarly implausible, each solving a difficult modeling problem, and each with different results), or effects estimated with econometrics that non-specialists have a lot of difficulty evaluating.  Moreover, we have to recognize that there is a selection bias in the results that get publicized or become part of the conventional wisdom in economics.  Economists value “clean” over “messy,” “innovative” over “standard” and “name” over “who?”  Moreover, much of the research on this question will come from Fed researchers, who have perhaps unconscious biases in deciding what lines of inquiry to pursue.

The Wall Street Journal has a nice roundup of research conclusions.  The roundup suggests maybe there were significant and sizable effect for mortgage rates, but not much else.  Daniel Thornton of the St. Louis Fed finds no effects of the QE program on yields or other outcomes.  Engen, Laubach,and Reifschneider find modest effects of the program (perhaps a change in unemployment of one percentage point), given its magnitude.  Here is a recent (and inconclusive) discussion by Jim Hamilton of Econbrowser (and an older discussion here) with lots of links.  Overall, as George Bernard Shaw (I think?) quipped, “You can lay all the economists end to end and never reach a conclusion.”