This post describes the mechanics behind quantitative tightening (“QT”) and reviews the prior QT experience. In our two-tiered monetary system, it is helpful to view QT through a framework that takes into account the perspectives of Banks (those who have a Fed account) and Non-Banks (those who don’t have a Fed account). Mechanically, QT reduces the level of cash held by Banks (reserves) and changes the composition of money held by Non-Banks (more Treasuries and fewer bank deposits). The Fed is unsure how low Bank reserve levels can fall before impacting the financial system, so it executes QT at a measured monthly pace. The prior QT experiment began in late 2017 and ended in September 2019, when a sudden spike in repo rates panicked the Fed into restarting quantitative easing.
Quantitative Tightening is when the Fed receives principal repayments from its Treasury holdings but does not roll them over into newly issued Treasuries. Instead, the Fed takes the proceeds and simply extinguishes them. The reserves disappear from the banking system with a few keyboard strokes just as they appeared with a few strokes during QE. The mechanical impact of these actions can be described through the eyes of the Fed, Treasury, Banks, and Non-Banks.
Fed: QT shrinks the Fed’s balance sheet. The Treasury security asset is repaid with the repayment extinguishing the Reserves liabilities created to purchase the Treasury in the first place. The QE transaction is reversed (see here for a QE Step-By-Step piece).
Treasury: Nothing changes except the ownership of Treasuries issued. Fed owns fewer Treasuries while Non-Banks own more. Note that the Treasury may alter its issuance patters during QT – such as issuing more short-term debt since longer dated debt may become more expensive in the absence of QE. QT can thus indirectly impact curve shape.
Banks: QT shrinks the collective balance sheet of banks by reducing their reserve assets and deposit liabilities. (Note that Banks can also purchase Treasuries, in which case they would simply swap reserve assets for Treasury securities).
Non-Banks: Non banks essentially swap one form of money (bank deposits) for another (Treasuries). But the two types of money are not perfectly substitutable: Treasuries are free of credit risk and offer some return, while bank deposits come with some credit risk and no return.
QT effectively increases the supply of Treasuries to the private sector in addition to new supply from ongoing deficits.
Fed (Mostly) Sets the Pace
The Fed proceeds cautiously with QT because it does not know how QT will affect the financial system. Specifically, the Fed is reducing bank reserve balances without knowing the minimum level of reserves balances the financial system requires to function. In the past, the Fed began with a slow monthly pace of $6b a month of Treasuries that ramped up to $30b a month. It then slowed the pace in mid 2019 with the thought of ending QT in late 2019. Note the maximum amount of QT the Fed can conduct each month is limited by the amount of maturing principal it receives that month. The amount maturing varies significantly each month and is largely a function of past policy choices.
Maturing principal that is in excess of the QT pace is rolled over at auction. For example: if the Fed receives $20b in maturing principal but is capping the pace of QT at $15b a month, then it will rollover $5b into next auction. The $5b will be rolled over in proportion to the offering sizes being auctioned. For example: if $40b of 3-year and $60b of 10-year were being offered, then $2b would be rolled into the 3-year (40%) and $3b would be rolled into the 10-year (60%). (For detailed rollover mechanics see here).
Minimum Required Reserve Levels
QT gradually drained bank reserve balances (and the RRP) until it was ended late 2019 after a sudden spike in repo rates. The Fed frequently surveyed banks throughout QT to gauge their minimum level of required reserves. Banks widely reported that they had a lot more reserves than they needed. A rough estimate based on the surveys suggested that the banking system as a whole needed about $600b in reserves, which implied that QT could’ve continued all to way to 2021.
In September 2019 repo rates exploded higher and panicked the Fed into thinking that QT had broken the market. Some policy makers viewed the spike as evidence that the level of reserve balances in the banking system had hit a minimum. The Fed promptly began to both lend in the repo market and restart QE in an effort to add reserves into the system. The repo market quickly stabilized, but it was difficult to tell if the spike was actually caused by an insufficient level of reserve balances. In any case, the repo spike marked the end of the Fed’s QT experiment.