This is the conclusion of a St Louis Fed paper:
There is substantial empirical evidence showing that QE can have beneficial real effects on the economy even at the ZLB. Yet, at the same time, showing theoretically that QE can be welfare improving has been difficult to achieve. Our study is motivated by the desire to reconcile these observations. For this reason, we construct a New Monetarist model characterized by aggregate and idiosyncratic demand shocks. Agents are heterogeneous with respect to the idiosyncratic shock, so that in every period some agents are more patient than others. This heterogeneity generates a distribution in asset holdings. We then study the optimal stabilization response of the central bank to demand shocks that hit the economy.
We find that several results hold in this environment. First, the ZLB can be the best interest rate policy the central bank can implement although that is not necessarily the case due to a price externality. However, even when the ZLB is the best policy, not all agents are satiated at the Friedman rule and therefore there is scope for central bank policies of liquidity provision. Second, we study a particular form of QE whereby the central bank purchases private debt via repo arrangements in response to demand shocks. We find that such a policy is welfare improving even at the ZLB since it can relax the liquidity constraint of impatient agents without harming the patient ones. We show this is true regardless of whether we have inflation or deflation at the ZLB. Third, due to a pricing externality, QE can be welfare improving for patient agents even if they are unconstrained at the ZLB.
Keep in mind though, what Buttonwood in the Economist wrote a couple of days ago:
The problem is that not all collateral is treated equally. Lenders worry that, if the borrower fails to repay, the securities they are left holding may not sell for their face value. So they apply a discount, or “haircut”, to the collateral, depending on its perceived riskiness. At times of stress, lenders get nervous and apply bigger discounts than before. This is what happened during the financial crisis (see table).
Bigger haircuts mean that borrowers need more collateral than before in order to fund themselves. “When market volatility jumps, funding capacity drops in tandem and often substantially,” writes Mr Howell. The result, a liquidity squeeze at the worst possible moment, is a template of how the next crisis may occur (although regulators are trying to reduce banks’ reliance on short-term funding).
Viewed in this light, global liquidity should not be measured merely by the size of central banks’ balance-sheets but by the availability of acceptable collateral as well. On Mr Howell’s calculations, global collateral shot up in the aftermath of the financial crisis, but grew much more slowly from 2012 onwards. This may explain why global growth has been so sluggish.
Traditional quantitative easing may do little to help. “Simply expanding the central bank balance-sheet by buying in Treasuries from the private sector is robbing Peter to pay Paul,” writes Mr Howell, since the bonds could have anyway been used as collateral for repo transactions.
Given that funding conditions resemble those in Victorian times, Mr Howell thinks that central banks should return to the policies favoured by Walter Bagehot, a former editor of this newspaper, and focus, above all, on the smooth running of the credit markets. If they do not, the risk is that a shortage of collateral may induce another funding squeeze; low as they are, government-bond yields may then fall even further as banks scramble to get hold of them for funding purposes.