Why the Cars Never Left

If you want to understand how a central bank cools an economy, you do not need a model. You need one sentence, and you can test it against your own behavior.

If I can park my money at 5.2 percent and do nothing, why would I risk the drive for 6?

That is the entire transmission mechanism, stated the way a person actually experiences it. Everything else — the dot plot, the balance sheet, the press conference — is machinery in service of that one question, asked simultaneously by every household, business, and institution that holds a dollar.

The formal name for the number in that sentence is the hurdle rate: the return an investment must clear before it is worth making. It is built from two pieces. The first is whatever you can earn taking no risk at all. The second is the premium you demand for accepting risk on top of it. Raise the first piece and the whole hurdle rises with it, instantly, for every project on earth.

This is why the Fed can cool an entire economy by moving a single overnight rate. It is not managing the economy directly. It is raising the bar that every private decision has to clear, and then waiting.

Picture a hundred projects lined up by expected return — a warehouse at 5.5 percent, apartments at 6.5, a data center at 8.5, a biotech at 13. Draw a horizontal line at the hurdle. Everything above the line gets built. Everything below it gets shelved, not because it became a bad idea, but because doing nothing became a better one. Move the line up two points and you have just cancelled a slice of the future economy without passing a law, printing a form, or telling anyone which projects to kill.

That slice is the cooling.

Here is the part that surprises students: the Federal Reserve does not set the interest rate on your mortgage, your car loan, your business line of credit, or the Treasury bill you might buy. It sets a target range for one thing — the rate at which banks lend reserves to each other overnight — and then uses its own tools to keep the market rate inside that range.

As of August 2026 that target range is 3.50 to 3.75 percent, and the effective federal funds rate is trading at 3.63 percent. That is the lever. That is all of it.

Everything else happens because that single overnight rate is the short end of a curve that every other price of money is hung from. The Fed names five channels through which the lever reaches you.

Note what is missing from that list. The Fed’s own explanation of how monetary policy works never mentions the money supply and never mentions velocity. Its causal chain runs from the policy rate to market rates to asset prices, credit and the dollar, and from there to spending. We will come back to why that absence matters.

Between March 2022 and July 2023 the Fed raised its target range eleven times, 525 basis points in sixteen months, from effectively zero to 5.25–5.50 percent. Then it sat there for fourteen months.

Suddenly a Treasury bill paid more than five percent. So did a money market fund, which is mostly a wrapper around Treasury bills and repo. For the first time in over a decade, doing nothing paid real money.

The Fed studied this itself and gave the reason plainly: money funds pass rising rates through to savers faster and more completely than banks do. Banks raise deposit rates slowly and incompletely, because they can. So when the policy rate jumps, money funds win the race and the money moves.

Look at the last four bars. That is the finding nobody expected, and it is the reason this piece exists.

The standard story says that when the Fed starts cutting, the cash on the sidelines comes back out and chases risk again. It did not happen. The Fed cut three times in 2025 and money market assets kept climbing — up $1.45 trillion since the first cut, with retail money growing faster than institutional money. Ordinary savers, having finally been shown what cash can pay, declined to give it back.

The cars never left the lot.

Money parking is not itself the cooling. It is the symptom. The cooling is what fails to happen on the other side of the hurdle — and the cleanest place to watch it is the market where capital most obviously chooses risk over safety.

High-yield issuance is the sound of companies borrowing to do risky things. In 2021, with cash paying nothing, it hit a record $488 billion. In 2022 it collapsed 77 percent, to $112 billion. That is not a market malfunction. That is the hurdle rate doing exactly what it is supposed to do.

The same pattern shows up everywhere you look for it. Business lending went flat for eighteen months — commercial and industrial loans at US banks were $2.80 trillion in December 2022 and still $2.76 trillion in June 2024. Housing starts fell 31 percent from their April 2022 peak. Initial public offerings fell 94 percent by proceeds between 2021 and 2022, from $153.6 billion to $8.5 billion, on the SIFMA definition that excludes blank-check companies.

Cooling does not look like things going wrong. It looks like things not getting built.

And that is precisely what makes it politically difficult. The recession you prevent is invisible. The factory that is not built, the company that does not go public, the hiring plan quietly shelved in March — none of it shows up as an event. It shows up years later as an unemployment rate, and by then nobody remembers the decision that caused it.

Somewhere in this lesson a student always asks about the velocity of money, usually because they have seen the equation. It deserves an honest answer rather than a confident one.

The intuition is real and useful. A dollar that sits in a money market fund is a dollar not bidding for a house, a truck, a machine, or an employee. If enough dollars sit still, the total volume of bidding falls, and prices stop rising as fast. That intuition is the reason the parking-lot framing works, and I would not throw it away.

But here is what you must know before you use the word in public.

Velocity is not measured. It is computed.

The M2 velocity series published by the Federal Reserve Bank of St. Louis is defined as nominal GDP divided by the M2 money stock. Nothing observes how fast dollars change hands. In the second quarter of 2026, nominal GDP was $32.48 trillion and M2 averaged $23.00 trillion. Divide one by the other and you get 1.412, which is exactly the published figure. There is no independent measurement anywhere in that number.

Which means the famous equation — money times velocity equals prices times output — is an accounting identity, true by construction, not a discovery about the world. It becomes a theory only if you additionally assume velocity is stable. It has not been stable in the United States since the early 1990s. M2 velocity ran near 2.0 before the 2008 crisis, collapsed to an all-time low of 1.126 in the second quarter of 2020, and has clawed back only about half of that, to 1.412 today.

So when someone says “inflation stayed low because velocity fell,” they have said something circular. It is arithmetically identical to saying nominal GDP grew more slowly than the money supply, which is just the data restated. It explains nothing.

The trap that catches everyone

There is a worse version of this error and it is worth learning by name. If you chart M1 velocity across 2020, you see it fall off a cliff — from 5.27 to 1.59 in a single quarter. It looks like the most dramatic economic event in modern history.

It is not an economic event at all. In April 2020 the Federal Reserve Board removed the six-transfer limit on savings deposits, and savings accounts were reclassified into M1, adding roughly $11 trillion to the denominator overnight. The velocity of M1 fell because M1 was redefined. Anyone who has published that chart as an economic finding has published a bookkeeping change.

Use velocity as an intuition pump. Use it as descriptive bookkeeping. Do not use it as a cause, and never use M1 velocity across 2020.

Everything described so far would be straightforward if it happened quickly. It does not. Milton Friedman gave the problem its permanent name in 1961: monetary policy operates with long and variable lags.

Most people know the conventional figure of twelve to eighteen months. That figure is roughly right for activity — output, hiring, investment. It is badly wrong for prices. Recent work by Aruoba and Drechsel, using disaggregated price indices, finds that the response of the aggregate PCE price index to a policy tightening does not turn significantly negative until more than three years have passed.

Sit with what that means operationally. The Fed is steering a vehicle whose steering wheel responds three years after you turn it, using a rear-view mirror that shows the road as it was two months ago. Every meeting is a bet about a world that does not exist yet.

This is why central bankers sound evasive. They are not hiding a model. They genuinely do not know, and neither does anyone else, whether the tightening they did in 2023 is still working its way through the price level today.

The received version of the last four years goes like this: the Fed hiked, inflation came down, the Fed is now normalizing, and the cash on the sidelines will eventually come back out.

Three of those four are wrong as of today, and a student who learns the received version will be confused by every headline they read.

The Fed is not normalizing. It cut three times in 2025, the last in December, and has not moved since. That is five consecutive holds. At the July 2026 meeting the vote was nine to three, and all three dissenters wanted a hike.

Inflation is not solved. This one requires care, because the answer depends entirely on which measure you name. Core CPI is 2.5 percent. Core PCE — the family of measures the Fed actually targets — is 3.3 percent, with headline PCE at 3.7. Those two facts are both true, they are eight tenths of a point apart on the core, and their relationship is inverted from its historical norm. Anyone who says “inflation is at 2.5 percent” without naming the measure is not being precise enough to be right.

And the cash did not come back out. That is Figure 01, and it is the most interesting fact in this entire piece.

Four years of tightening taught American savers what cash is worth. They have not agreed to unlearn it.

Which gives the parking-lot metaphor a sharper ending than the textbook version. The lesson is not simply that high rates make money park. It is that once money has learned where the good parking is, lowering the rate a little does not send it back onto the road. Behavior, once changed, has its own inertia — and that inertia is not in anybody’s model.

Already true

The target range has sat at 3.50–3.75 percent since December 2025 with three dissents for a hike in July. Core PCE is 3.3 percent against a 2 percent target. Money market fund assets reached a record $8.29 trillion in the first quarter of 2026, up 21 percent since the first cut.

What has to happen

For the standard story to be right, core PCE has to resume falling toward 2 percent while the Fed holds, and money has to eventually leave the risk-free option for risk. Neither is happening yet. If both fail, the Fed hikes again, and every hurdle rate in the country rises with it.

Where I am probably wrong

I may be over-reading the money market data. Assets can grow simply because the pool of savings grows and because yields compound inside the funds, not because anyone made a fresh decision to stay parked. And the 2026 rise in headline inflation is heavily energy — the energy index is up 14.7 percent year over year — which is precisely the kind of shock the Fed is supposed to look through rather than fight. If both of those are the whole story, the received narrative is closer to right than I am giving it credit for.

1. The Fed raises the bar, it does not pick the projects. One overnight rate lifts every hurdle rate in the country simultaneously, and the market decides which ideas die.
2. Cooling is an absence, not an event. It looks like deals that do not close and factories that do not get built. That invisibility is exactly why the job is politically brutal.
3. Always name the inflation measure. Core CPI at 2.5 percent and core PCE at 3.3 percent are simultaneously true. Only one of them is the target.
4. Velocity is computed, not observed. Use it as intuition, never as a cause, and never chart M1 velocity across 2020.
5. The lag on prices is measured in years, not quarters. Twelve to eighteen months describes activity. The best current estimate for the price level is over three years.
6. Behavior has memory. Money that learned what cash pays does not automatically forget when the rate comes down a point.

The through-line connecting this piece to the last one is a single idea, and it is worth stating plainly one more time. There is always a number that represents the reward for doing nothing. Every price in the economy — a park, a bond, a factory, a hiring plan — is quoted relative to it. The Federal Reserve moves that number.

Which is why a student who understands the spread understands more of the financial world than a student who has memorized a hundred cap rates.