I do have one slight criticism. Sumner argues:
While the demand for base money by the public, that is, the demand for currency, doesn't usually change very rapidly, if very short term interest rates fall to zero, and the interest rate on reserve balances at the Fed is negative, then what "usually" happens is irrelevant. A flight to safety--to T-bills, FDIC insured deposits, and balances in reserve deposits at the Fed--should be cleared up by higher prices and/or lower yields on those particular assets. But what if a negative yield is necessary? What if those desiring these safe and liquid assets are willing to pay to hold them and the issuers require payment to issue them in sufficient quantities? What prevents this from happening is that safe and zero-yielding currency is no longer inferior to these assets in terms of its yield, but superior. The demand for currency will rise to match the excess demand for the other assets at a yield equal to the storage cost of currency.
In the very unlikely event that all these excess reserves were then hoarded by the public, the Fed could have begun quantitative easing in October 2008. I very much doubt this last step would have been necessary, as demand for base money by the public (in the absence of 1930s-style bank runs), does not change very rapidly.
My view is that once Lehman failed, and especially after the stock market crashed, substantial quantitative easing, including open market operations using longer term and riskier assets than T-bills, was necessary. Paradoxically, as Sumner often argues, the mere expectation that the Fed was willing to undertake such radical steps could have made them unnecessary. The problems in short term credit markets created by the flight to safety were at least partially, and perhaps mostly, caused by the Fed's failure to act aggressively and promptly during the summer of 2008.