Scott Sumner made a very simple argument. He had read something by Sheldon Richman suggesting that monetary policy benefits those who receive the new money first, because they are able to spend the new money before prices rise.
Sumner responded with three scenarios. In the first scenario, the Fed purchases Treasury bonds from a bank. In the second, the Fed purchases the Treasury bonds from a securities dealer. And in the third, the Fed purchases the Treasury bonds directly from the Treasury which uses the money to pay the salaries of Federal workers. (To make Federal workers the literal first receivers of the money he described it in an odd way. The Fed pays the workers, and the Treasury gives the Treasury bond to the Fed as a gift.)
It is important to understand that in the first two scenarios, the Fed is purchasing the exact same asset. If the price of these Treasury bonds should rise because of the Fed's purchase, then the benefit of a slight capital gain goes to both the bank and the securities dealer. There is no special gain to the one from whom the Fed actually makes the purchase.
Now, the scenario where the Federal workers are paid is a bit more complicated. The Treasury would have sold bonds on the market to pay the workers. This might have slightly depressed bond prices, and by having the Fed buy the bonds instead bond prices are maintained. If the Fed had instead purchased the bonds from the bank or the securities dealer, the Treasury would have sold the bonds to fund the salaries. But the purchase of the bonds from the bank or the securities dealer just offsets the sale of securities by the Treasury.
Sumner's conclusion is that regardless of whether the Fed purchases the Treasury bonds from the Treasury to pay the workers, from the bank, or from the securities dealer, the Treasury bond prices are the same and the Federal workers still get paid. The Federal workers don't especially benefit if the Fed purchases Treasury bonds to pay them instead of the Treasury selling the bonds on the market and the Fed purchases the Treasury bonds from the bank or the securities dealer. And the security dealer doesn't benefit if the Fed purchases his bonds rather than those of the bank or else buys them directly from the Treasury. And the same for the bank.
Now, Sumner did not say that it makes no difference if the Fed purchases Treasury bonds or Mortgage-backed securities. He said that makes a little difference, but not no difference. If the Fed is buying a single homogeneous asset and it is paying the market price, it doesn't matter from whom it buys that asset. And Richman seemed to suggest it does matter--those who get the new money first supposedly benefit.
Where is Richman going wrong? Suppose there is an economy using gold for money and the quantity of money is 1000 ounces. A gold miner discovers 100 ounces of gold. He announces his good fortune, and everyone immediately raises their prices 10 percent. The gold miner isn't spending the money before prices rise. Prices already rose. But the gold miner certainly is better off. He is able to purchase goods and services that are now worth 100 ounces of gold, but that were before worth 90 ounces of gold. Still, there is no doubt that the gold miner benefits from finding the gold. Even if prices rise immediately, he gets 90% of the gain.
If the gold miner's new discovery was only 10 ounces of gold, and prices immediately rose one percent, then the gold miner would buy 10 ounces worth of goods, but goods that were worth approximately 9.9 ounces of gold before. He would gain 99% of the benefit.
Of course, if the gold miner kept his discovery secret and spent his gold before prices rose, then he buys 10 ounces worth of goods, which is what 10 ounces purchased before. The point, however, is that the gain from spending the money before prices rise is insignificant. Finding the gold did benefit the gold miner, allows him to purchase goods close to the previous value of the gold, and any benefit from spending the newly found gold before prices rise is trivial.
A counterfeiter is much like the gold miner. The gain isn't that he spends the money before prices rise. If the quantity of money is $1000 and the counterfeiter creates $100, and prices did rise 10% instantly, even before he spent the money, he would be able to buy $100 worth of goods, what $90 would have purchased before. He obtains 90% of the gain, even if prices rose before he spent the money he created. If he had only created $10, he would get 99% of the benefit. Yes, spending money before prices rose would be better, but it is not the chief source of the gain.
If we consider the government itself to be like the counterfeiter, and it issues fiat currency to purchase some particular good, then even if prices rose immediately due to the increase in the quantity of money, it can still purchase some of the good--just a small amount less than it could have purchased if prices had not increased. It is almost certain that the price of that particular good would rise relative to the prices of other goods. And further, it is certain that the allocation of resources would shift to produce more of the good purchased by the government with the nearly-created money.
Where do the resources come from for the gold miner, the counterfeiter, or the government to obtain these goods? All of those currently using the money must reduce their real expenditures in order to increase their nominal money holdings so that their real money holdings return to the desired level. This reduction of real expenditures below real income for those paying the "inflation tax" frees up resources to produce the goods purchased by gold miners, the counterfeiter, or the government.
What is the difference between finding gold or printing currency to spend and an open market operation? The securities dealer and the bank exchange a Treasury bond for the money. The Federal workers, on the other hand, do seem closer to the government simply printing up money and purchasing some particular good. But Sumner set up the scenario so that they would have been paid anyway. The only question is whether some private investor buys the Treasury bonds to pay them and the Fed purchasesTreasury bonds from the bank or securities dealer or else the Fed buys the Treasury bonds itself to pay the workers. They are paid the exact same.
In my view, Richman's (and other Austro-populists') argument that the "elites" profit from monetary policy by getting the new money first, is wrong. It confuses the gain that goes to someone creating (or finding) money and spending it with someone who sells something in exchange for newly created money.
However, just because there is no special benefit from being the one getting the new money from the Fed in exchange for some good or asset as opposed to selling the same asset (or good) for "old" money, doesn't mean that there are no "injection effects." As I have discussed before, it is important to distinguish the consequences of alternative trend inflation rates and the consequences of an excess supply of (or demand for) money. Various trend inflation rates most likely result in slightly different allocations of resources, but these different allocations can be persistent. On the other hand, excess supplies of money cannot be persistent and they can impact the allocation of resrouces, but it doubtful whether an ephemeral excess supply of money has any significant impact on the allocation of resources.