Scott Sumner argued against the notion that because the federal government can print money, it doesn't need to worry about the cost of providing services. He argued that while the issue of currency allows the government to collect a small revenue--perhaps 1% of GDP--through seignorge, government spending is much greater than that, so on the margin, changes in government spending require taxes.
In response to a comment, Sumner then amended his argument to recognize that seignorage revenue is not free. However, he argued that if we compare a scenario with no currency to one with currency, those bearing the cost of holding currency receive benefits greater than that cost.
As an analogy, he describes the introduction of a government lottery. Those participating in the lottery get to enjoy legal gambling, and the government collects the revenue. He notes that since private competitive lotteries are clearly possible, a more appropriate baseline shows that funds from a government monopoly lottery are not "free" but rather they come at the expense of those purchasing lottery tickets.
Sumner then considers privatized hand-to-hand currency and repeats his longstanding proposal to have the government contract out the provision of hand-to-hand currency. He holds that competitive issue would be wasteful, pointing to the practice of banks of giving away toasters to those making deposits as a work-around interest rate ceilings.
First, the seignorage income of the government from the monopoly issue of currency is imposed on those using currency. With a zero inflation rate and growing currency demand, those wishing to increase real currency balances must increase their nominal currency balances. They do that by having nominal expenditures less than nominal receipts. Real consumption plus real saving in forms other than the accumulation of money balances must be less than real income.
With inflation, say at the 2% rate Sumner favors, it is necessary to accumulate nominal balances to maintain real balances. Again, nominal expenditures must exceed nominal receipts. And so, real consumption plus real saving must be less than real income. The inflation tax on real money balances is paid by those seeking to maintain their real money balances.
Evidently, people benefit from holding currency more than this cost. But the tax on currency balances results in an excess burden like any other tax. The conventional wisdom is that the proper baseline for comparison is a deflation rate equal to the real rate of interest so that there is no opportunity cost to holding currency. Little seignorage revenue is possible with a deflation rate equal to the real interest rate. If the real interest rate is equal to the growth rate of real output and the demand for currency is proportional to real income, then there is exactly no seignorage revenue--a constant nominal quantity of money is optimal.
Just about all of this analysis is worthless when considering free banking. Banknotes are debt instruments. While the nominal interest rate may be zero, a bank that issues them is still borrowing and must stand ready to pay them all back. There is nothing like seignorage revenue for any bank.
Of course, borrowing at a zero-nominal interest rate is lucrative under normal circumstances. Banks can fund a portion of their asset portfolios with no interest cost. The immediate effect then is to enhance the profitability of banks.
However, if banks make more than normal profits, then there will be entry into the industry. Considering banks as financial intermediaries, the resulting increase supply of banking services will reduce the equilibrium margin between the interest rates banks charge and pay. Entry continues until profitability returns to normal. The impact then will be an increase in the interest rates banks pay on deposits and a decrease in the interest rates banks charge on loans.
And so, if the government monopolizes the issue of currency so that it can collect seignorage, then this comes at the expense the customers of banks--you and I. We earn lower interest on bank deposits and pay higher interest on bank loans.
Hand-to-hand paper currency was initially issued by private, competing banks. The government step-by-step monopolized the issue through legal restrictions. Since this paper currency was initially redeemable in terms of gold, it was either a liability of a private central bank or a type of government debt By creating a monopoly, the tendency of competition to dissipate the rents made possible from borrowing at a zero nominal interest rate could be shared by the private owners of the central bank and the government As time passed, those benefits have gone more and more to governments. With the end of the gold standard, it became possible to think of this monopoly government currency as if it is paper gold--a pure outside money with the amount issued creating a revenue.
But the private alternative remains a competitive banking system. My own view is that it is entirely possible to pay interest on hand-to-hand currency. When depositors withdraw currency, they can continue to earn interest until the currency is returned to their bank. One hundred years ago, the record keeping would have been very burdensome, but it is very feasible today.
Further, to the degree it is desirable to tie price level performance to optimal holdings of hand-to-hand currency, how much more likely would this occur when it doesn't involve government giving up a source of "free" revenue and instead involves shifts in how the benefits are distributed among the customers of banks?