Market participants have coalesced around a September start to the tapering of quantitative easing. There are, however, a few notable exceptions such as Bill McBride at Calculated Risk and the economics team at Bank of America Merrill Lynch, both wary that the data will support such a policy shift. That puts me on the opposite side of a bet with McBride, which is something that tends to make me nervous. Kind of like the idea of playing Russian Roulette with five chambers loaded.
That said, I think September is the date to begin tapering, and the data flow would need to turn notably downward to forestall a policy shift at that time. I think it is important to recognize that the Federal Reserve is treating quantitative easing and interest rates as two very separate policies, and each has its own relevant data. From the BAML report cited by McBride:
Although the Fed has attempted to clarify its reaction function, we have become increasingly uncertain. The FOMC has zeroed in on the jobs market and to a lesser extent in reduced downside risks: both argue strongly for near-term tapering. However, the Fed's mandate is to manage the overall economy and the gap between solid jobs and weakness in other growth and inflation indicators has gotten very big...
Exactly -- quantitative easing has always been primarily about the job market and mitigating downside risks, essentially putting a floor under the economy. Indeed, the decision to initiate open-ended quantitative easing was made on the back of a bad jobs report that was subsequently revised substantially upward; I suspect that many policymakers have some buyers remorse, recognizing that without that preliminary data release they wouldn't be struggling with the tapering issue now.
Likewise, I suspect the conversion of Operation Twist to an outright purchase program was largely about addressing potential downside risks from fiscal policy. Now such risk looks minimal, and consequently the additional asset purchases look unnecessary.
Finally, while there has been weakness in the growth and inflation indicators, it has already been entirely dismissed by policymakers. (See my discussion of recent speeches by New York Federal Reserve President William Dudley.) Indeed, the solid jobs report for June will only feed their idea that other data should be downplayed with regard to quantitative easing. The same, however, is not true for interest rate policy, hence why the Federal Reserve is more committed to a longer period of near-zero interest rates than might have been expected given the improvement in the data they cite as reasons to scale back asset purchases.
More to the point, however, is that they are not entirely comfortable with quantitative easing and want to quickly bring the program to a conclusion. Hence the bar to ending quantitative easing is relative low now. They are more comfortable with zero interest rates, and thus have a relatively high bar for changing interest rates.
In short, to accept a September tapering as a data dependent decision, you need to accept that the data threshold is relatively low and differs from the threshold for interest rate policy. They are two separate policies. Consequently, we don't need to see substantially better data to forestall a September taper, but instead substantially worse data.
BAML raises the possibly more important issue the pace of tapering:
Three scenarios seem plausible. (1) The rest of the economy quickly converges to the employment data and the Fed starts a steady move to the exit in September. (2) The Fed decides reduced downside risks make the case for a one-time dial down in QE, so they taper in September but then pause for an extended period waiting for clear broad-based improvement. In other words, subsequent moves are more data dependent than the first. (3) The Fed decides to wait for broad confirmation in data and doesn't start tapering until December. The third option remains our base case, but clearly we are out of the consensus and September tapering is increasingly possible.
I have already argued against options 1 and 3. Option 2 is still open and fits well with the position outlined by Federal Reserve Governor Jeremy Stein, namely that as we approach the September meeting, weak economic data will be considered less important for the outcome of that meeting relative to subsequent meetings. Barring a substantial negative turn in the data, the Fed will cut the pace of asset purchases in September, but a softer data flow may reduce the magnitude of the cut and the pace of subsequent tapering.
That said, given the communication challenges the Fed faces, my guess is that they will want to avoid Option 2 and will use the upward trend of non-farm payrolls to do so. If they come in with a small cut in September, they may set the stage for substantial financial market volatility if they then proceed with larger cuts in subsequent meetings. They won't want to be the source of such volatility. Hence, that's why I suspect they are already on a pretty much calendar-dependent path -- such is the easiest path to manage market expectations. Once they begin, I expect they will trim asset purchases by roughly $15-$20 billion a meeting until they reach zero by the middle of next year.
Bottom line: The Federal Reserve is having a difficult time convincing market participants that quantitative easing and interest rates represent two separate policy tools. They want to severe the perception that the two are connected -- a reduction in the pace of asset purchases thus does not signal a change in the expected lift-off from the zero bound. Understanding that the two policies are different is, I think, key to understanding why the Fed is heading toward a September tapering, despite what many view as an overall subpar economic environment.