The U.S. Federal Reserve’s efforts to quell inflation have sent long-term interest rates to their highest level in a generation, putting a lot of stress on banks, companies and anyone looking to finance a new home.

How long will this go on? Judging from the sheer volume of long-term debt securities that the Fed still needs to unload, I’d say at least a couple more years.

For more than a year, the Fed has been shrinking its holdings of Treasuries and mortgage-backed securities at a rate of about $75 billion a month. This process, known as quantitative tightening, gradually reduces the amount of excess cash reserves in the banking system. The aim is to reach a level of reserves adequate to ensure that banks can satisfy their customers’ variable demands for cash and meet regulatory liquidity requirements.

The last time the Fed embarked on quantitative tightening, in 2018 and 2019, it went a bit too far. The supply of reserves fell below what banks required, triggering a scramble for cash that sent short-term interest rates spiking and destabilized the repo market, where banks, hedge funds and others borrow money against Treasury and mortgage-backed securities. The Fed was forced to step in with emergency liquidity—an experience that it doesn’t want to repeat.

This time around, the Fed’s starting point is much higher. Its holdings of long-term debt securities exceed $7 trillion, compared with $4 trillion in 2018. Reserves in the banking system stand at about 12% of gross domestic product, up from 7% in September 2019. If one assumes—consistent with the most recent report on the Fed’s open market operations—that 8% of GDP would be a suitable cushion of reserves, and that the Fed won’t slow the rate of quantitative tightening until reserves fall to 10% of GDP, there’s still a long way to go.

The shrinkage is likely to proceed even if the economy slows and the Fed needs to ease. The federal funds rate is its primary tool of monetary policy: It employs the balance sheet only when short-term rates are pinned near the zero lower bound. With the target rate currently above 5%, the central bank has plenty of room for maneuver without resorting to altering the path of quantitative tightening. So it should proceed on autopilot—about as exciting as “watching paint dry,” as Janet Yellen once remarked.

There’s also a new complicating factor: the Fed’s reverse repo facility, where money market mutual funds and others have parked some $1.5 trillion (as of October 11), at interest rates exceeding 5%. As quantitative tightening continues, cash will migrate from the Fed’s facility to other repo markets in pursuit of higher rates, increasing the supply of reserves. (If this doesn’t happen on its own, the Fed will help it along by slightly lowering the reverse repo rate relative to other money market rates.) About $1 trillion has already moved over the past year, even as the Fed has shrunk its balance sheet by $800 billion, resulting in a net addition of $200 billion in reserves.

So when will quantitative tightening end? Assuming an annual runoff rate of $900 billion, nominal GDP growth of 4% and reverse repo balances declining to zero, reserves should reach the initial target of 10% of GDP in about two years. At that point, the Fed will slow the run-off pace as it assesses what constitutes an appropriate reserve buffer.

I see three main repercussions. The first will be upward pressure on long-term interest rates and on the bond term premium—the added yield investors demand to lend for longer. Second, higher term premiums will tighten financial conditions, allowing the Fed to keep interest rates lower than it otherwise would (as Dallas Fed President Lorie Logan implied in a recent speech).

Finally, by increasing the volume of securities that market participants must absorb and finance, quantitative tightening will risk further turbulence in the Treasury market. Currently, the supply of Treasuries is expanding by nearly 10% of GDP ($900 billion in Fed runoff plus a federal budget deficit of $1.7 trillion), potentially straining the balance sheets of the dealers at the center of the market.

It’s possible that serious dysfunction in the Treasury market could cause the Fed to relent. Barring that, quantitative tightening should proceed unabated, and this should put upward pressure on long-term interest rates until at least late 2025.

Bill Dudley, a Bloomberg Opinion columnist and senior advisor to Bloomberg Economics, is a senior research scholar at Princeton University’s Center for Economic Policy Studies. He served as president of the Federal Reserve Bank of New York from 2009 to 2018, and as vice chairman of the Federal Open Market Committee. He was previously chief U.S. economist at Goldman Sachs. He has been a nonexecutive director at Swiss bank UBS since 2019.