The big message I believe everyone is missing about the bond market is, "It's the duration mismatch, stupid."
As Mish has explained in past posts, lenders borrow at low rates for short terms and lend at higher rates for longer terms. Lenders have to roll over their short term loans until their customers repay longer term loans. Lenders pocket the difference between the yields. By realizing the existence of a disinflationary environment and that yield curves rarely invert; lenders overcommitted to a perceived low risk / high reward trade. It's not the "free money" of an arbitrage, but pretty close.
As declining inflation takes hold, yields come down. Spreads between loan durations shrink. To hit profit targets, lenders have to take greater duration risks. Instead of lending for 10 years financed with 5 year loans rolled over once, lenders might lend for 10 years by financing with a 2 year loan rolled over 4 times. This places more demand on short term lending; resulting in short term rates rising relative to long term rates. Eventually lenders reach the point of financing long term using overnight lending.
Once the yield curve inverts at the short end, one could ask, "Why not borrow long term at lower rates and lend short term at higher rates?" Initially that does start to happen. Some lenders are willing to finance several sequential overnight loans with a single one month loan. Since there is a statistical risk of an inverted yield curves lasting a short period of time, lenders are willing to take a small duration mismatch in the other direction. Some of those rolling over loans begin to close their duration mismatch a little more aggressively by financing with longer short term loans. Now demand increases at the 2 month duration, which morphs into longer and longer durations. Bond markets are pretty much completely inverted for all duations. Duration mismatch can no longer be used to maximize profits. Too many lenders still have loans to roll over, but no one to finance them. Bank reserves are insufficient to repay the short term loans and still meet reserve requirements. Liquidity has dried up to the point US bills and Treasuries can't be sold on the open market to avoid defaults.
In late summer 2019 the Federal Reserve stepped into the repo market "to provide liquidity" by purchasing US bills and Treasuries. If the Federal Reserve had not stepped in, many short term loans lenders took out would have defaulted, creating an insolvency chain reaction. The fact that the Federal Reserve has publicly stated its operations will last for months is a testament to severe structural imbalances duration mismatch has created.
This has lead to a scholarly debate of "Not Quantitative Easing" vs "QE". I'll take the "Not QE" side of the argument. Based on the above analysis, "Not QE" is a euphamism for "insolvency crisis". Given the inverted yield curve, companies and home buyers already have every incentive to borrow large amounts of money at historically low costs; so there are no capital constraints holding back the economy.
None of the Federal Reserve's "Not QE" efforts can fix the duration mismatch alone. A change in banking regulations is required to prohibit duration mismatch. It won't matter if such a regulation is phased in slowly or instantly because investors decide on how fast changes in regulation will affect their investments. The amount of short term loans needing to be rolled over also effects how quickly policy change will impact portfolios. Yields would necessarily have to rise once duration matching starts. This would crash both the stock and bond markets.
The Federal Reserve has several short term options to kick the can down the road. It is already stepping into the repo market. Overnight repo can become the remainder of the bill's or Treasury note's duration. This is a limited solution because lenders have a limited supply of US bills and Treasuries in their portfolio. The Federal Reserve could reduce banks' reserve requirements. This is also limited.
The Federal Reserve really doesn't have any long term tools with positive outcomes. The Federal Reserve can lower it's overnight rate in an attempt to normalize the yield curve from the short end. The drawback is retail depositors would have to pay storage fees instead of receive dividends. Selling long term US bonds to normalize the yield curve would harm long term economic growth. Adding corporate debt to acceptable repo assets would require a rule change. Such a move should create a fear that markets would no longer be truly free, but at least the perception of hyperinflation can be avoided. None of the above solutions discourages duration mismatch, the risk that has finally destabilized the