The hypothetical QE3, whatever its format, might not work so well because the last two rounds were so successful.
That’s what Goldman analysts Sven Jari Stehn and colleagues say.
They’ve measured previous QEs by their effect on Goldman’s Financial Conditions Index, which they say eased by about 100 basis points per $1,000bn of asset purchases. Specifically, this is how the first two rounds measured up by this model:
Specifically, we explain the level of the GSFCI with three components. First, we include the announced stock of the Fed’s asset purchases (that is the announcements in November 2008, March 2009 and November 2011 to purchase $600bn, $1.15trn and $600bn of securities, respectively). Second, we include the target fed funds rate (as a measure of the current stance of conventional monetary policy) and the slope of the Eurodollar curve (to capture expectations for future monetary policy).
Finally, we include a number of economic variables that capture other economic influences, including Reuters/University of Michigan long-term inflation expectations and initial jobless claims. We estimate this model using weekly data between January 2000 and August 2011. We find that the first two rounds of asset purchases, on average, eased financial conditions by 101bp per $1trn (see column 1 in the table below). This estimate is consistent with our previous range of estimates, but a bit higher than their average.
However they say things might not work so well next time around — because, well, things are not so bad:
There is, however, reason to believe that QE3 might be less effective in easing financial conditions than the first two rounds of purchases. This is because Fed asset purchases are likely to have larger effects during times of extreme market stress, and particularly when they are targeted at a specific market dislocation, like the mortgage-related purchases during QE1.
As a result, one would expect that QE1 had a more significant effect on financial conditions than the subsequent program. Unfortunately, it is difficult to test for this empirically as we only have two experiences to go by. That said, we find some tentative evidence that the effect of QE1 on financial conditions was larger than during QE2: the effect rises to 120bp per $1trn when we estimate the model only through the end of QE1 in March 2010 (see column 2 above). To the extent that this finding points to diminishing returns to quantitative easing, we might expect additional purchases to ease financial conditions by less than 100bp per $1trn of purchases (or an equivalent extension of the average duration of the balance sheet).
Another drawback for QE3, they say, is that in a near-zero funds rate environment, asset purchases have the effect similar to shortening the average maturity of government debt — however the average maturity of privately-held Treasury debt has actually been increasing.
As for the real economy? There too, QE rounds past have had some effect:
When it comes to QE3, however, it is seen as being less effective on GDP than previous rounds — mainly due to the housing market’s effect on consumption, and energy prices:
In particular, a significantly positive effect on housing construction seems unlikely given the large amount of unoccupied inventory that currently hangs over that market. Moreover, the response of consumption might well be more muted in light of households’ inability to extract equity from homes through refinancing due to widespread negative equity. In a simple attempt to capture these effects, we fully exclude the contribution of residential investment to GDP growth implied from the chart. (An alternative approach would have been to exclude parts of both the housing and consumption response.) Doing so suggests that a 100bp easing in financial conditions might boost growth by only 0.5 percentage point in the first year (see dashed line). Second, it is possible that rising energy prices might offset some of the growth effects of any QE3-induced easing in financial conditions.
In conclusion: QE3 will be no panacea. But then we’ve been pointing that out for a while.