The most troubling aspect of his paper is that he finds "quantitative policy," that is, changes in the quantity of base money, to be irrelevant for inflation And while the size of the Fed's balance sheet has no impact on inflation (and presumably nominal expenditure and any demand-constrained production,) he quite paradoxically sees benefits in an indeterminately large level of bank reserves.
How does he come to such odd conclusions?
He asserts that the Fed's actual policy focus is on changes in the Federal Funds rate. Certainly, there is no doubt that the FOMC targets the Federal Funds rate.
He then develops a model with two interest rates--a short term rate which the Fed is controlling and the interest rate the Fed pays on reserves. Aggregate demand depends on this short term interest rate, adjusted for expected inflation. And the inflation rate depends on aggregate demand. Of course, there is more to it, but that is the basic chain of causation, and it is certainly the conventional approach of modern macro.
He argues, plausibly enough, that if the Fed keeps the short term interest rate at the level that will keep inflation on the desired target, then nothing else is going to matter for inflation. It is a bit odd that one can follow this approach while discussing nominal interest rates being at the "zero bound," but other than that, the model can be solved showing that the Fed can vary short term interest rates in a way that keeps inflation on target.
But this whole approach is based on the premise that the Fed likes targeting short term interest rates. What if that changes? Surely, that is what is at issue. What happens if the Fed instead targets the size of its balance sheet and lets the Federal Funds rate vary?
Reis' puzzling result is that the size of the Fed's balance sheet must at least be at the level of "satiation," but can be any size larger than that. He goes so far as to argue that this would be a good thing, because it will allow the Fed to provide as much "liquidity" as needed.
Reis' result is driven by his simple model. While there are two interest rates, the odd result comes from reducing that to one interest rate. That rate is set at the level that provides macroeconomic stability. And the Fed pays that same rate on reserve balances. Reserve balances are then a perfect substitute for both government and private securities (both of which pay the single interest rate that the Fed sets.) Satiation means that any extra benefits that might be provided by holding reserve balances rather than bonds are obtained. After that point, the Fed can hold the bonds and let the public hold the interest bearing reserves, or it can let the public hold the bonds.
Why would anyone hold any private bonds if the interest rate on them was the same as the interest rate on reserves, and reserves are perfectly liquid and risk free? Why would anyone hold any government bonds? While they presumably have the same risk as reserve balances, what possible benefit would their lack of liquidity provide? A one interest rate model just breaks down.
It is true, of course, that by paying interest on reserves, the Fed can increase the demand to hold reserves. By adjusting that interest rate, it can manipulate the demand for them.
It has always been true that it is the quantity of reserves in combination with the demand to hold them that creates their macroeconomic impact.
The Fed could leave the quantity of reserves constant, and implement a monetary policy by changing the interest rate paid on reserves and so varying the real demand for them. In the short run, this can impact market interest rates and real output. In the long run, it impacts the price level and the real quantity of reserves.
Or, the Fed could change the interest rate it pays on reserves to increase real demand while increasing nominal quantity to accommodate that demand. And so, the Fed can choose to expand the size of its balance sheet in a way that remains consistent with macroeconomic equilibrium.
And, of course, this will enhance the ability of the Fed to manipulate the allocation of credit in the economy. Further, the Fed can bear risk and provide liquidity to the private sector--holding risky and illiquid securities and providing riskless and perfectly liquid reserve balances. While providing these services has some costs, the Fed could bear those costs. It could subsidize the provision of riskless and liquid assets.
What are we to make of Reis' extreme case where there is one interest rate? The extreme case where the Fed pays "the interest rate" on reserves? This is just exploring the case where the subsidy is complete and the Fed is willing to bear all of the risk and bear all the cost of providing liquidity.
While Reis mentions that his model assumes that "satiation" occurs at a less than infinite quantity of reserves, isn't the most likely scenario that the Fed allocates all credit? That the quantity of money would be equal to total wealth? That the demand for money would also be total wealth? That the interest rate paid on that money, and the interest rate changed for that credit would need to be the natural interest rate that keeps saving and investment equal?
Fortunately, there is not just one interest rate, and the Fed can seek to cover the cost of bearing risk and providing liquidity by paying a lower interest rate than it earns on its asset portfolio. And maybe, the Fed should avoid trying to manipulate credit, and hold low risk and liquid assets. Perhaps the Fed should just stick to stabilizing the growth path of nominal expenditure, and let the market allocate credit.