He asserts that growth of nominal GDP should be the goal of monetary policy (and links to an article where he defends nominal GDP targeting.)
He claims that the proper target and indicator for monetary policy is the price of a futures contract on nominal GDP.
He favors using both the level of base money and the interest rate paid on base money (in the form of bank reserves) as instruments.
Base money should grow at roughly the target rate of nominal GDP (adjusted to the price of index futures contract on nominal GDP on target), and then the interest rate on reserves should be relegated to keeping the total quantity of base money equal to an appropriate share of nominal GDP.
In my opinion, the proper goal for a monetary regime is to provide the least bad macroeconomic environment for microeconomic coordination. At first pass, the regime should generate a quantity of money that adjusts to the demand to hold money. However, the demand to hold money depends on the way the unit of account is defined. So, I would add that a monetary regime should result in planned aggregate spending on goods and services growing at a rate consistent with the long run trend growth rate of the productive capacity of the economy.
This is certainly more complicated than growth of nominal GDP. However, I would translate my proposed goal for a monetary regime into a specific target rule of a 3% growth path for nominal GDP. Sumner doesn't specify a growth rate at all and I think he should have done more to distinguish between a growth rate and growth path "goal."
What is the difference between between the "goal" for a monetary regime and the target rule? It involves several questions. For example, should nominal GDP be the measure for spending on output? Or should final sales or final purchases be used instead? And, while the goal I outlined specifies the growth rate of spending on output in general terms--equal to the trend growth rate of productive capacity--exactly what that rate would be is part of the target rule.
Like Sumner, I am very interested in using index futures contracts. My approach focuses on the central bank's position on the contract rather than the market price of the contract, but I am not sure that it would be an adequate indicator or target. I see it more as a side constraint. Those who trade these contracts must be looking at something else as an indicator of monetary policy. Even if the central bank always acted to keep its position on the contract at zero (which is similar to keeping the price of the contract on some kind of target,) it would really be adjusting current monetary conditions according to whatever indicators and intermediate targets those trading the contracts were using.
Also, I believe that the central bank should be free to adjust monetary conditions as it sees fit, and should never have to wait until others trade a futures contract. If no one bothers to trade the contract, then the central bank directs monetary policy, subject to the target rule. The role of the contract, in my view, is to provide signals and incentives to the central bank to make adjustments according to the trades of the contract by speculators. Still, that means that there must be some other indicator or target that the central bank would be using. Determine the stance of monetary policy it sees as best, and then choose whether or not to make adjustments according to the expectations of the private sector as revealed by trades of the contract.
In my view, the fundamental goal of the monetary regime is to adjust the quantity of money to the demand to hold it. That includes a central bank issuing monetary liabilities, so adjusting the quantity of base money seems like the right policy instrument to me.
Unlike Sumner, I think paying interest on reserves is an excellent idea, however, I do not see efforts by the central bank to optimize the amount of liquidity held in aggregate to be desirable. On the contrary, the primary benefit of paying interest on reserves is to allow it to passively change with other interest rates. That way, changes in other interest rates do not impact the opportunity cost of holding reserves and so the demand for them. That interest rates paid on reserve balances increase the demand for reserves and so the size of the central bank's balance sheet, and particularly the amount of assets it must hold, is a necessary evil.
Most money pays interest. Banks borrow short and lend long and so bear risk. This risk is a reason why the interest rate that can be earned on money is lower than the interest that can be earned by lending longer. The banks' stockholders are bearing this risk for those holding the money.
This risk is real. Investment projects take time to bring to fruition, and those holding money don't have to wait to spend the money. Sumner's proposal is to have the central bank compete with ordinary banks in this business, borrowing short by issuing money and lending long. Why should a central bank be expected to do a better job? Surely, a central bank can subsidize this risk. But will real world central banks improve matters? Why exactly is discouraging people from making commitments to wait until real investment projects mature a good thing?
Of course, a decrease in the demand for the money issued by a private bank can easily be matched by an increase in the demand for base money. The solution to this problem is to have the central bank adjust the quantity of base money it issues according to the demand. There should never be a want for liquidity because there is "too little" central bank money.
Does this mean that the central bank is immune from this problem? No. There can be a decrease in the demand for base money and a central bank committed to a target rule for nominal GDP is bound to respond. It can pay higher interest on base money or else sell off assets from its portfolio. That is exactly what any bank must do, and so the central bank is subject to losses in exactly the same way. The risk of loss is on the central bank's stockholders (the government?) rather than the private owners of an ordinary bank.
The only difference is that if there is a decease in the demand for the money issued by commercial banks and an increase in the demand for base money, and the quantity of base money fails to increase, then the resulting deflation in the prices of assets, reductions in output and prices, and shifts of wealth from debtors to creditors can destroy private banks. If, on the other hand, there is a decrease in the demand for base money, and the central bank fails to decrease the quantity of base money or raise the interest rate paid on it, then the impact is inflationary.
This is not a reason for a central bank to try to provide "liquidity" by paying interest on reserves. The interest rate on reserves should not be used as a "tool" to encourage banks to stay "liquid" while blowing up the central bank's asset portfolio. As an alternative, I would suggest organizing reserve balances more like a money market mutual fund invested in T-bills. Any gains or losses would be passed through to the banks, with the central bank charging a management fee.
A growth path for Nominal GDP as the target rule.
Index Futures contracts as the policy constraint.
The quantity of base money as the policy instrument.