Goldman Expects $10-15bn Taper And Fed Walking-Back From Employment Thresholds

With bonds and stocks rallying (and the USD dropping) notably in the last few days, one could be forgiven for believing the Taper is off but Goldman’s baseline forecast remains for a $10bn reduction in asset purchases – probably all in Treasuries – and $15bn is possible (though recently mixed labor data may choke that a little) and a strengthening of forward-guidance. As they note, the current redction in uncertainty (or rise in complacency some might say) has the potential to offset the tightening in financial conditions, barring another major outbreak of DC strife in the run up to the debt ceiling in late October/early November. However, what is most notable is Goldman’s expectation that the Fed will start walking-back its unemployment-rate threshold as it has been clearly shown not to be a good catch-all indicator of broad economic and labor market performance. So it’s data-dependent – but the data is unreliable at best and false at worst.

Via Goldman Sachs’ Jan Hatzius,

1. Growth seems to have picked up a bit in August, and our current activity indicator (CAI) is now just shy of 3%. It is true that the August employment report was not great, with a slightly lower-than-expected gain in nonfarm payrolls and a decline in the employment/population ratio. But we recently showed that the outsized market response to the preliminary nonfarm payroll number each month bears little relationship to its limited actual value for figuring out the growth of the economy. In contrast, the jobless claims and ISM numbers have been quite good, and they are more useful indicators than payrolls in our view.

2. Our baseline forecast is a gradual improvement in growth as the drag from fiscal policy declines late this year and in 2014. One downside risk to this is the tightening in financial conditions, and specifically the 100bp increase in mortgage rates. Our GSFCI stands about 30bp tighter than prior to Chairman Bernanke’s May 22 testimony, a number that, if sustained, could potentially shave ¼-½pp off growth over the next year, with most of that impact coming from housing.

3. But we are not yet too concerned. Admittedly, the recent housing numbers have been on the weaker side but over the medium term our confidence in the basic math of recovery remains high. We believe that the trend rate of household formation will average at least 1.2 million over the next few years. Add to this an estimated 300,000 demolitions of existing homes per year, and the homebuilding industry will need to produce 1.5 million new homes per year to keep excess supply stable. In the near term, some further reductions in excess supply are likely, but the 2-3 year outlook is still a big increase in housing starts from the current 900k level.

4. The reduction in various measures of uncertainty since late 2012 is another reason for a measure of optimism. We remain skeptical of the view that the weakness of the recovery from the 2007-2009 recession was primarily due to policy uncertainty because we believe that much of the increase in uncertainty over that period was more the result than the cause of economic weakness. But the recent broad-based drop in uncertainty looks more clearly exogenous, and therefore deserves more weight in an economic forecast. Our analysis suggests that the reduction in uncertainty has the potential to offset the tightening in financial conditions, barring another major outbreak of DC strife in the run up to the debt ceiling in late October/early November.

5. Although growth may be starting to improve, we still see a strong case for accommodative monetary policy as core PCE inflation stands at 1.2% and employment remains far below potential. Admittedly, the renewed drop in labor force participation over the past two months is a challenge to our expectation of a flattening out of the participation rate in coming years. But the case for looking beyond the unemployment rate does not rest on the cyclical/structural decomposition of the participation rate alone. Even if we focus on more observable measures such as U6—the Labor Department’s broadest definition of underemployment, which also includes marginally attached workers and involuntary part-timers—we find that there is substantially more slack in the labor market than suggested by the unemployment rate alone.

6. Fed officials will take all of this disparate information into account when they announce their September 18 decision. Our baseline assumption remains a tapering of QE, although we have shaved the amount to just $10bn—probably all in Treasuries—in view of the mixed labor market data and recent signals that Fed officials are uncomfortable with a large step (although $15bn is certainly possible). More importantly, we expect them to offset the potential impact of tapering via a strengthening of the forward guidance. Our baseline is that the statement will make the 6.5% unemployment threshold conditional on a return of inflation to the 2% target and will indicate even more explicitly that continued below-target inflation would translate into a longer lag between reaching the 6.5% threshold and the first hike. We also expect Fed officials to further increase their emphasis on the importance of broad labor market improvement. This may involve an outright statement that they would only consider the threshold to be met if the decline in unemployment reflects an increase in employment as opposed a drop in participation.

7. A more aggressive approach would be to simply reduce the unemployment threshold from 6.5% to 6%. We believe the probability of such a move is significant. The rationale is that the July FOMC minutes indicate a widespread view that the drop in unemployment since they adopted the 6.5% target overstates the improvement in broad labor market conditions, which may call for a recalibration of the 6.5% number. Moreover, from a more tactical perspective an outright reduction in the threshold is a very “crisp” way of pushing back against market expectations of an early hike. Nevertheless, such a move is not quite our baseline. Especially if the taper is small, a reduction in the threshold would not only constitute a change in the mix of instruments but an outright easing. This may be more aggressive than the committee is willing to be at a time of generally slightly better economic data, uncertainty about the identity of the next Fed chair, and the inevitable concerns about the credibility of forward guidance associated with that uncertainty. However, if they decide to taper QE by more than $10bn, the probability for an outright threshold reduction would be higher in our view.

8. The other big question is how Chairman Bernanke will revise his forward guidance for the path of asset purchases in the press conference. In June, he said that purchases would likely end around mid-2014, a time when the FOMC expected the unemployment rate to be around 7%. Assuming the committee’s unemployment forecast has come down—after all, the unemployment rate stood at 7.28% in the August report—he will have to revise either the mid-2014 date or the 7% unemployment rate. Our expectation is the latter, although this may take the form of increased wiggle room as opposed to a new number.

9. Stepping back, we believe that Fed officials probably suffer from some degree of “buyer’s remorse” with respect to the unemployment thresholds that they introduced formally for the first rate hike in December 2012 and informally for the end of QE in June 2013. Although the conceptual argument for guiding expectations in terms of economic outcomes as opposed to the calendar is sound, we noted at the time that this only works well in practice if the unemployment rate is a good catch-all indicator of broad economic and labor market performance. Unfortunately, this has not been the case recently, and Fed officials are likely to respond by trying to loosen the shackles that they have imposed on themselves.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.