The answer seems to be: growth.
Donald Luskin at Trend Macrolytics has estimated that every $4bn (£2.48bn) of QE generates $1bn of share buying: the other $3bn is recycled into cash and other, generally safe, assets. This implies that if the Fed were to end its current $85bn per month of QE - which it will not do suddenly - then demand for equities will fall by around $21bn per month. This is only 0.1 per cent of the US's total market capitalisation, as measured by the Wilshire 5000, the broadest index. This implies that a cessation of QE would take around 1.2 per cent off of US share prices over 12 months.
How does this compare to the effect of growth? A simple regression equation tells us that since 1996 each percentage point of above-average growth in industrial production has been associated with 2.9 percentage points above-average returns on the S&P 500. Variations in output growth alone explain almost half the variation in S&P returns in this period.
Putting these two estimates together tells us that the benefits to share prices of output growth being only 0.4 percentage points above-average would offset the damage done by an ending of QE. This is not much. It suggests that growth matters more than QE.
This conclusion is consistent with research by the New York Fed's Carlo Rosa. He's estimated that $500bn of QE has similar effects on asset prices to a half-point cut in the fed funds rate, which implies that the ending of $1bn of QE over 12 months would have a similar effect to a one percentage point rise in the funds rate. This is the sort of move markets can usually cope with. For example, between June 2004 and June 2006, the fed funds rate rose by four percentage points, and yet the S&P 500 rose more than 10 per cent.
It's also consistent with the market's reaction to the last employment report. This showed job creation to be stronger than expected. But shares rose on this news, even though it raised the prospect of the Fed tapering sooner than expected. This suggests growth trumps QE.
It's also consistent with rough intuition. If we get stronger than expected growth, where could investors put their money? Not into cash, because interest rates will stay low for some time. And not into bonds either, because growth is bad for these - a fact likely to be exacerbated by the withdrawal of QE. And commodities and emerging markets are unlikely to benefit much, simply because these are also sensitive to the withdrawal of QE.
This implies that if the US economy grows sufficiently well for the Fed to withdraw QE faster than the market expects, then share prices should rise because the good growth news will be stronger than the bad monetary policy news. Pessimists, on the other hand, might infer that if the market gets some bad news about growth then the prospect of continued QE won't be sufficient to protect share prices.
What, then, might be wrong with this verdict? One possibility is that QE doesn’t help shares merely by encouraging investors to put cash into the market. It also offers insurance against downside risk generally - the so-called 'Bernanke put.' Without this insurance, shares will be riskier and so their prices should be lower.
I'm not sure about this. For one thing, better economic growth should also reduce some kinds of downside risk. And for another, withdrawing QE needn’t be irreversible; the Fed can resume it if necessary.
Another possibility is that my estimate of the effect of growth upon share prices is too high. It could be that the strong correlation between growth and equity returns doesn't just reflect the fact that growth raises share prices, but the opposite - that higher share prices cause faster real growth, via wealth effects upon consumer spending or because a lower cost of capital encourages investment.
However, this reverse causality is probably weak. For one thing, the fact that capital spending isn't financed by equity issues tells us that higher share prices do little to raise investment. And for another, wealth effects aren't big. A recent study by Yale University's Robert Shiller and Wellesley College's Karl Case concluded that there is "at best weak evidence of a link between stock market wealth and consumption".
I suspect, then, that the correlation between economic growth and equity returns is due very largely to the former causing the latter - for example by increasing investors' appetite for risk. This implies that growth matters more than QE.
There are, though, two ways in which the threat of reduced QE might hurt equities.
One would be if the prospect of tapering emerges without offsetting good news on growth. This would happen if inflation rises unexpectedly, or if the Fed looks like being more hawkish than the market expects. It’s this that caused shares to fall (sharply but temporarily) in May and June.
A second danger would be if the market extrapolates from a small initial tapering of QE that policy will tighten further in future - with QE being ended or even reversed and the fed funds rising. Such a normalisation of monetary policy must happen eventually. The question is how expectations of it enter prices. And this depends upon the solution to a coordination game; each individual investor is trying to anticipate when and whether others will worry about future monetary tightening. It’s probably impossible to know the answer to this in advance.
On balance, it's probably the case that, over the longer-run, share prices will be driven more by economic growth than by the withdrawal of QE - although there'll probably be brief periods when the opposite is the case. In this sense, investors should worry less about when QE will be reduced, and more about whether the US economy can continue to grow.
