I recently argued that banks do not play an important role in monetary policy. Rather we need to focus on the supply and demand for base money, which is mostly determined by central bank policies. The blog Spontaneous Finance recently commented as follows.
You print more currency than the public wants to hold, and they'll bid up prices. How do you inject it without banks? Simple, pay government worker salaries in cash. Or buy T-bonds for cash. Cantillon effects don't matter, unless the central bank is doing something bizarre, like buying bananas.
This is an unrealistic story. Does any central bank print money and pay government salaries? No. At least not in the developed world. From whom do central banks buy T-bond? From...banks, and from funds, which then leave the newly acquired cash in...banks. Or which purchase other assets to replace those T-bonds, in which case this cash also ends up in banks. And in truth, the hot potato effect described by Leland Yeager can easily get interrupted by the operational realities of the financial sector. In short, financial institutions can wear heat protection gloves. There is no need for IOR for excess reserves to build up. The implementation of monetary policy remains subject to strong structural rigidities (it isn't the goal of this post to list such rigidities, although it may be the topic of a later one).
Against all evidence, Sumner keeps denying that banks, their business models, their regulation, and their accounting standards, play a role in transmitting monetary policy, booms and busts. He (as well as Hetzel) considers himself a follower of Milton Friedman's monetarism. Yet Friedman seemed to understand more about banks than Sumner does. Indeed, Friedman and Schwartz partly blamed bank accounting standards (in particular mark-to-market accounting) for the catastrophic banking collapse (and money supply collapse) of the Great Depression.
I completely agree with Friedman and Schwartz that banking played a major role in the Great Depression. But that's because we were mostly on the gold standard, and the price level (and NGDP) at that time was determined by changes in the supply and demand for gold. Central banks had almost no direct influence on the supply of gold, but did have some influence on gold demand.
The same was true of the banking system. A wave of bank failures in the US, combined with near zero interest rates, led to a surge in demand for bank reserves. These reserves had to be backed by the government gold stock (or at least 40% backed), and hence banking turmoil led to more central bank demand for gold. It also led to more private demand for gold (called hoarding.) This all had a deflationary impact, as gold gained purchasing power. As soon as FDR started devaluing the dollar in April 1933, gold demand no longer held back the recovery and prices started rising. The banking system became a non-issue until the dollar was re-pegged to gold in January 1934.
Under a fiat money regime the central bank has unlimited control over the supply of base money, so changes in commercial bank demand for reserves are not an important factor, unless the central bank has some inhibition about expanding its balance sheet during a period where they have (foolishly) allowed interest rates to fall to zero. Or, even worse, paid interest on reserves.
One other point. Prior to 2008, an injection of money by the Fed went overwhelming into cash, not the banking system. What did go into banks was mostly immobilized as required reserves. But over 97% of new money created by the Fed between August 2000 and August 2008 went into cash held by the public, of which the vast majority was $100 bills. Prior to 2008, the Fed steered NGDP mostly by controlling the supply of "Benjamins."
Now it's true that the new money initially went into armored cars, and then banks, before spilling out into cash in circulation. But that detour took just a few days at most, and had no important macroeconomic consequences.
Between August 2007 and May 2008 the Fed sharply slowed the rate at which it injected $100 bills into the economy. We all know what happened as a result of currency growth slowing to zero:
PS. It's actually sort of unusual for a recession to be triggered by a sharp slowdown in the rate of growth in currency---usually it's caused by the Fed failing to accommodate a rise in the demand for currency.
PPS. I know that central banks do not give newly created money to public employees---my point was that it wouldn't matter if they did.
PPPS. Here's the data for August 2008, (NSA)
Bank deposits at Fed = $10.454 billion
Vault cash used for required reserves = $35.337 billion
Vault cash not used as reserves = $17.123 billion
Cash held by the public = $775.500 billion
That means the currency stock was $827.960 billion and deposits at the Fed were barely 1% of the money created by the Fed. The rest was currency (held by the public or banks.) Unless you understand monetary policy prior to 2008, you can't really understand the subject.