The DXY (U.S. dollar index) may be pulled by two opposing forces but the outcome will be the same: a weaker dollar. If the US rebound is confirmed by the NFP on Friday, Sept. 7th, the "risk-on" move initiated by a better US news flow and the ECB's bond purchase program announcement should continue. Otherwise, any sign of an insufficient recovery could trigger a more accommodative monetary policy, which could be seen as a bearish signal for the USD.
Risk-On: A better news flow combined with higher risky asset prices is clearly the "worst" outcome for the DXY. The disconnect between the VIX (risk aversion) and the DXY seems to be over. Of course, VIX is already low, but a return to its recent trough would be consistent with a DXY close to its 61.8 Fibonacci retracement of 80.70 (which is, incidentally, also the current level of 200-day moving average).
Risk-Off: a lot has been written on Jackson Hole and the likelihood of a QE3. The US news flow may have improved, but the question of the sustainability of the growth momentum remains. With an ECB unwilling to cut rates (0.75% for the repo rate) and a possible dovish "forward guidance" from the Fed, the Dollar may take a hit.
The main uncertainty pertains to the impact of a potential QE, a new bond purchase program, on 10-year yield. I would lean towards a more flexible program with no pre-fixed amount, as suggested by Williams (The Fed would, in a sense, be inspired by the ECB's Outright Monetary Transactions program).
Any form of QE could have an impact on 10-year yield, but the final outcome is clearly uncertain: QE1, QE2 and Operation Twist had significantly different impacts on the direction on US Treasuries (see chart below). The most efficient operation was, as can be seen below, Operation Twist, with a significant cap on yields and a disconnect with the news flow. The main explanation may come from the fact that it was balance-sheet neutral.
Another point that is worth mentioning: the disconnect between the News Flow and long term rates is almost over. A new QE would try to accompany a weak recovery, not a faltering growth (I expect 2% US GDP growth for H2). It would look like early QE2. Hence, the direction of long rates would probably be slightly up.
Lastly, the chart below shows that the correlation between the DXY and 10-year UST yields has turned structurally negative. A new QE would be have a negative impact on the USD, only if it drove rates upward - which is for US, the most likely outcome.
The odds for a weaker DXY are quite high:
1. in a risk-on environment, the correlation between VIX and DXY has returned to positive territory: hence a likely lower USD;
2. In a risk off environment, dovish forward guidance from the Fed would weaken the USD whereas a "flexible QE" would be DXY negative (as can be seen above).
Now that the SP 500 has reached a 4-year high, it would probably come with some favorable arbitrage towards emerging market (see below).