Is the interest rate on reserves holding the economy back?

Martin Feldstein claims to have solved the puzzle of why Quantitative Easing has not resulted in higher inflation:

The link between bond purchases and the money stock depends on the role of commercial banks’ “excess reserves.” When the Fed buys Treasury bonds or other assets like mortgage-backed securities, it creates “reserves” for the commercial banks, which the banks deposit at the Fed itself... 

The link between Fed bond purchases and the subsequent growth of the money stock changed after 2008, because the Fed began to pay interest on excess reserves. The interest rate on these totally safe and liquid deposits induced the banks to maintain excess reserves at the Fed instead of lending and creating deposits to absorb the increased reserves, as they would have done before 2008. 

As a result, the volume of excess reserves held at the Fed increased dramatically from less than $2 billion in 2008 to $1.8 trillion now. But the new Fed policy of paying interest on excess reserves meant that this increased availability of excess reserves did not lead after 2008 to much faster deposit growth and a much larger stock of money. 

So it is not surprising that inflation has remained so moderate – indeed, lower than in any decade since the end of World War II. And it is also not surprising that quantitative easing has done so little to increase nominal spending and real economic activity.

This explanation seems highly dubious to me.

Econ 102 says that banks lend money to long-term risky projects. They choose to lend if the expected real rate of return from a risky project is greater than r + S, where r is the safe real rate of return, and S is some required spread.

With IROR > T-bill rate (as now), the IROR is the safe asset. With IROR < T-bill rate, the T-bill rate is the safe asset.

The IROR is 0.25%. The T-bill rate is just over 0%. This means that the difference in the expected real rate of return between a world with an IROR and a world without an IROR is about 0.25%. In a world without an IROR, banks lend to any risky project with an expected real rate of return of S. In a world with an IROR, banks lend to any risky project with an expected real rate of return of S + 0.25%.

Therefore, what Feldstein is asserting is that there is an absolutely huge number of risky projects whose expected return is between S and S+0.25. He is asserting that if we lowered the safe rate by 0.25%, a huge panoply of projects would then become worth investing in, and a huge torrential flood of reserves would be released into the economy, boosting inflation and lowering unemployment in the process.

Does that seem reasonable? To me, that does not seem reasonable in the slightest. First of all, it is a priori dubious, because of the small numbers involved. Second, since S actually differs from bank to bank, depending on lots of different factors, it makes sense that if Feldstein were right, then some of the big banks would be lending like gangbusters and others wouldn't. Is this what we see? To my knowledge, it is not.

Therefore I conclude that the IROR is not the reason why QE has not translated into higher bank lending, higher inflation, and lower unemployment. The answer must lie elsewhere; it must be the case that either S is very high, or there are very few projects with decent expected rates of return, or maybe some sort of institutional constraint on the speed with which banks can ramp up lending. But that little 0.25% IROR can't be what's holding the economy back.


Update: Also, as Matt Rognlie emails to remind me, Japan didn't pay interest on reserves from 2001 to 2008. And Japanese bank lending went nowhere during that period despite a burst of QE. So that is another nail in the coffin of the IROR hypothesis. Always remember to look at Japan!