Capitalist economies have historically seemed prone to large recessions. The Market Monetarist school of thought believes that this is not an inherent feature of capitalism, but instead a feature of our current regulatory system and monetary policy regime. Can the right monetary policy make the world safe for capitalism?
I like to listen to audiobooks to fall asleep, and I recently picked up Scott Sumner's The Money Illusion. This has turned out to be a mistake, since the book is so gripping it keeps me up until the morning light every time I listen to it.
I don't know much about macroeconomics, but it got me thinking and I think I've come up with a novel idea for how to dramatically improve the US banking system and prevent a large class of recessions from ever happening again.

What Causes (Demand) Recessions?

There's the classic helicopter-drop thought experiment, where you double the money supply by dropping a ton of cash from a helicopter. It stands to reason that if you attempt this, in the long run you won't have actually accomplished much - everyone will buy more stuff with all their new money, then prices will go up because everyone's buying so much stuff, then people start to demand more money from their jobs because everything just got more expensive and their company is making a lot more money, and then eventually everyone is making twice as much and everything costs as twice and really not much has changed except the number in everyone's bank account is twice as high.
You can do the reverse, a helicopter-suction thought experiment, where you halve the money supply, everyone has less money to buy things, prices go down, people get pay cuts, and in the long run everything is the same as before except the number in everyone's bank account is half as high.
You'd think that helicopter-suction would be symmetrical to helicopter-drop. There's a complication though, which is that prices are downwards sticky. What this means is that companies irrationally hate lowering prices and employees hate getting pay cuts. So, after you do the helicopter-suction, instead of just slashing everyone's wages by half, companies either go out of business or fire half their workforce or do anything except give everyone a 50% pay cut. This happens even if it doesn't change people's real income, i.e. their income after adjusting for inflation or deflation. If prices are halved, people are happy. But if that means your business now makes half as much money so you have to cut salaries by half, people will riot and your company will go out of business.
This is why deflation is so bad for an economy. In the late 1800s there was "the great deflation" where prices kept going down and companies kept having to fire people and it was basically just an extended economic catastrophe. So now central banks try pretty hard to prevent deflation, and once in a while they even succeed. In 2007 and 2008, there was some deflation, and this probably really contributed to the great recession.
There are two main causes of deflation:
  1. Rapid economic growth.
  2. A nominal reduction in demand, possibly caused by people having less money.
The toxic thing is, once you get a little bit of deflation, the natural tendency is for there to be more deflation. Once a company goes out of business, its employees have less money, which causes more deflation, which causes more companies to go out of business, etc.
Furthermore, if you can't pay back your loans, you default, which means the people who loaned you money might default too, which adds up to a viscous cycle where a bunch of money that people thought they had (because they loaned it to someone, who loaned it to someone, etc.) turns out to not really exist. Most money in the economy is loaned in some form, either because it's sitting in a bank account or used to invest in the stock market and so on. More on this later! Since this money "disappears", it leads to deflation, which causes even more businesses to fail, which means they can't repay their loans, which causes more money to disappear, which leads to more deflation, etc. It's a whole mess.
So central banks really want to prevent deflation from happening. But they're not very good at it!

Why Central Banks Suck at Fighting Deflation

Normally, central banks try to adjust the amount of money available to people by adjusting interest rates on loans. When you put money in the bank, the bank reserves some of the money to leave with the Fed - key word is "some", they only do that with some fraction of all the money people have in their bank accounts and the rest they can do whatever they want with. If you put $10 in a bank account and the reserve requirement is one half, banks have to leave half of your money with the Fed and with the other half they can do whatever they want. The fraction of bank deposits that banks must leave with the Fed is called the reserve requirement. They can leave more than the reserve requirement with the Fed if they want to, but not less.
If the reserve requirement is one half, and you make a bank deposit, that means the bank has to leave one half with the Fed, but they're free to take the other half of your money and loan it to someone to make money off interest. Let's say they loan it to someone looking to buy a house: The homebuyer borrows the money and gives it to the seller in exchange for the house, at which point the seller may deposit it back into the bank. The bank again leaves half of that money with the Fed and loans the other half to someone else, where it may end up in another bank account, etc. This means that combined balances in everyone's bank account can be much higher than the actual amount of base money the government created! So loans create a kind of "virtual money" that contributes to inflation all the same as base money.
The Fed has various techniques to control the amount of virtual money, and that's their main tool for controlling inflation. What are those tools?
The amount of virtual money is dependent on how many people have taken loans. How many people take loans is determined by the interest rate that banks charge for those loans. Higher interest rate = less money loaned = less virtual money created. Lower interest rate = more loans = more virtual money created. If the central bank can manage to lower interest rates, more people will borrow and there will be more inflation than there otherwise would be. One example of something the Fed can do to lower interest rates is to lower the reserve requirement.
Unfortunately, interest rates suffer from the problem of the "zero bound", which is that it doesn't really make sense to have banks give loans at a negative interest rate. That would mean the bank is paying you money to take a loan!
This sucks, because times when interest rates are low might be exactly the time your currency is most in need of inflation! Interest rates naturally become very low whenever there's deflation, and if the Fed is trying to lower interest rates when there's deflation and can't lower them past zero, that's an obvious problem.
(You can get banks to lend even more by charging them to leave money with the Fed, but there are various pitfalls with that. To prevent banks from acting irresponsibly, they have various requirements that they have to meet before they can lend money. An example is capital requirements that make it so they have to have a certain amount of capital before they can give new loans. In 2007 many banks became undercapitalized when the stock market plunged, which meant they were unable to give new loans unless they somehow got more money. Banks were acting irresponsibly already so the Fed didn't want to just lower the capital requirement, and they probably couldn't just cut all the banks a check because with the Occupy Wall Street movement there there was bipartisan disdain towards banks and the finance industry. Many think this is why the Fed picked the, uh, unintuitive strategy of paying banks interest on reserves, which is the exact opposite of what you'd think they'd want to do! By paying banks interest on reserves, banks had a risk-free way to make money without lending, so they actually lent less than before. It's possible that the idea was that by paying banks, banks would be able to get more capital, which plausibly could have allowed them to lend more in the future as they'd now have enough cash to meet the capital requirement. I'm not sure if this is actually what happened though.)

The Banking System Is Upside-Down

I propose the Fed creates what I'll call the Cash Account. This acts like a normal bank account from which you can deposit or withdraw money. For now let's assume the Cash Account pays no interest - it's just a digital equivalent to stuffing $100 bills under your mattress.
Having the Cash Account would give the Fed a fun power. Whenever they want more inflation, they can just deposit money into everyone's cash account. The natural way to do this is to multiply the balance of everyone's cash account by some number >1. So if they think there should be 2% more base money and all cash is stored in cash accounts, they can just multiply everyone's cash account balance by 1.02, and presto, the supply of base money just went up 2%. Currently, the Fed also issues bonds called T-bills that pay a small interest - in my system, the value of these would also be multiplied when the government multiplies Cash Accounts.
Let's assume this initiative is a success and paper cash is phased out. If the Fed wanted to cause deflation for some reason, they could use the same mechanism, except they choose a multiplier <1. This seems to me like a powerful approach to combating hyperinflation. And if people expect the Fed to actually use this capability if hyperinflation happens, it might mean hyperinflation never actually happens in the first place and they never have to use it.

Interest on cash accounts

I suggest 3 separate cash accounts - a risk-free account that can hold an unlimited amount of money and bears no interest, a risk-free account that bears interest but has a limit of $250k (each person can only have one account), and a risk-bearing interest-bearing account with no limit.
For the interest-bearing accounts, where does the interest come from? It's a simple 3 step process:
  1. Anyone can pay $10M/year and become a bank, to whom the Fed may eventually lend money from Cash Accounts. The amount the have to pay also scales up as the Fed lends them more.
  2. The Fed uses the $10M to set up a subsidized prediction market that predicts the chance the bank will default in the next 5 years. If they default, the Fed will be on the hook for all the money they promised was risk-free, so they'll use this probability to compute how much they need to charge for our equivalent of FDIC insurance. (The exact computation for the amount they have to pay is: the government's liability if the bank completely fails × the prediction market's opinion of the chance that the bank will default)
  3. The banks tell the government how much they're willing to borrow at every interest rate (a curve that they can update at any time). The Fed computes the risk-adjusted interest rate so that it includes the chance of default. Then the government simply lends the money from cash accounts to whichever bank is offering the highest risk-adjusted interest rate, provided the bank is willing to pay the cost of our bank-failure insurance. Every day following, banks make another insurance payment according to the prediction market's latest probabilities of bank failure.
    (If the bank does default, then some of the money in the interest-bearing risk-bearing accounts will disappear. That's life when you have more than $250k!)
Is this endeavor safe for the government? Yes. The money the banks pay upfront is a form of insurance, and as long as the insurance isn't underpriced, the government in expectation won't lose money having to reimburse people's risk-free accounts if the banks fail. Not only will the insurance not be underpriced, it'll be overpriced for two reasons:
  1. The bank may default after only being able to pay back, let's say, 98% of what they borrowed from the government. If this happens, the insurance was overpriced, because the the insurance assumed that when they defaulted they'd lose all of the government's money instead of only 98%.
  2. People will be tempted to buy "the bank will default" shares in the prediction market to hedge against the money in their risk-bearing account being lost. This adds a pressure on the market to be biased towards predicting that the bank will default, although the subsidization will offset this.
(It's possible that the first factor will cause the price of insurance to be dramatically higher than it should be. If this is a problem, the government can have the prediction market be one where predictors submit a probability distribution over the amounts the bank is expected not to be able to repay. But it's less obvious to me how to subsidize these so I hope this isn't a problem.)
This is basically recreating a credit union except it's run by the government. So the Cash Account interest rate should be competitive with the interest rate you can get from traditional bank accounts today.
Note: When the government multiplies the value of the cash accounts, the debt the banks have to the government is left unchanged. If the multiplication increased the money supply, the additional money is just loaned to the banks at their newest rates.
Banks that lend to average joes, like the ones of today, can continue to exist and be FDIC insured as usual. But since the money in the Cash Account will increase when the Fed increases the money supply, there should be a powerful demand for Cash Accounts over normal bank accounts.
This system has the pleasing property that if someone realizes that a bank will fail, they can buy a ton of "the bank will default" shares in the prediction market and make a huge profit, while at the same time causing the government lend to them less and increase the amount they must pay in liability insurance. Also, if just one bank fails, everyone will lose a fraction of the money in their risk-bearing cash accounts, which seems better than the current system where people who picked the wrong bank lose everything and others lose nothing.

When Should the Fed Increase the Money Supply?

Market monetarists have a bunch of good ideas here. Here's the short version:
Every year the government should figure out average American's labor income. The Fed's goal is path targeting such that the average income rises along a path of increasing about 4% each year. (Path targeting means that if you undershoot your target one year, the next year you raise your target to make up for the shortfall.)
There's a pretty straightforward way to do this: set up two bonds, one that pays out of the Fed undershoots their path, and another that pays out if the Fed overshoots their path. Then you set it up so every time a bond is bought that says the Fed will undershoot, there's an automatic increase in the money supply. And every time a bond is bought that says the Fed will overshoot, there's an automatic decrease in the money supply. Therefore, the money supply will always be set at a level that causes the market to believe the Fed will hit its target. What could be simpler?
Consider what happens if a recession starts. This will probably lower the average income, right? So to keep the average income on-path, the Fed will have to increase the money supply quite a lot - that will prevent the economy from sliding into deflation and the stimulus will heat up the economy to ease the recession. And instead of relying on the Fed, who may not react until the recession has already been going on for a long while (happens more often than you'd think), we only rely on the market, which can plausibly react much faster and incorporate more information than the Fed can.
Why might the market react more quickly? Academic economists are not necessarily very good at predicting the market. From The Money Illusion:
Because macroeconomic data often come out after a long delay and because it takes at least several observations to discern a new trend, economists often don’t even recognize that the US is in a recession until it is well under way. To noneconomists, this must sound absolutely appalling. Not only are we unable to forecast recessions, we can’t even “nowcast” them! During the past three recessions (prior to COVID-19), the consensus of professional economic forecasters didn’t predict a recession until roughly six months after the downturn had begun.
If academic economists were better at predicting recessions than the finance industry, the finance industry would pay them millions of dollars for their knowledge until there was no longer any difference. So the finance industry should be at least as good if not better than the academic economists who currently work at the Fed!
(Again, I don't take any credit for this part, this is all mainstream market monetarist stuff.)
Pathing against average income is the market monetarist opinion, but it's not clear to me why average income is the thing to set your path against. Why not median income, or some combination?