Markets

What the Federal Reserve Actually Does

The Fed targets one overnight rate between banks, not your mortgage. How a target range becomes a price, why it leaks on the way to you, and what the balance sheet changes.

By the Euphoria team · 2026-07-30 · 9 min read

The rate it sets is a rate you cannot get

Every few weeks a headline announces that the Federal Reserve raised, cut, or held interest rates, and a reader reasonably concludes that somebody in Washington adjusted the number on a mortgage or a savings account. Nothing like that happened.

The Fed sets a target for exactly one interest rate: what banks charge each other to borrow money overnight, unsecured, for a single night. You cannot borrow at it and your bank is under no obligation to pass it to you. It is not even one number. Since December 2008 the Federal Open Market Committee has announced a target range a quarter of a percentage point wide, and the rate that actually trades is whatever banks negotiate inside that band.

As of July 30, 2026 the range was 3.50 to 3.75 percent, where the Committee had left it since December 11, 2025, after a cutting cycle that took it down from a peak of 5.25 to 5.50 percent. You can check the whole history yourself on the Fed's own open market operations page. The effective federal funds rate, the volume weighted median of what banks actually paid, sat at 3.63 percent inside that range. The distance between the announcement and the transaction is where most of the confusion about this institution lives.

• 3.50% to 3.75% the federal funds target range on July 30, 2026
• 3.63% the rate banks actually paid each other inside that range
• 9 to 3 the July 2026 vote, with three members preferring a hike

How a target becomes a price

A target is a wish until something enforces it. What enforces this one is not a rule but an offer, and understanding that offer is the single most useful thing in this article.

The Fed pays interest on the money banks park in their accounts at the Fed. That payment is called interest on reserve balances, and in July 2026 it stood at 3.65 percent. A bank that can earn 3.65 percent risk free by doing nothing will not lend to another bank for much less, so that rate acts as a floor with a little slack in it.

Underneath sits a second offer, the overnight reverse repurchase agreement facility, which lets money market funds and other institutions that have no account at the Fed lend to it overnight against Treasury collateral. Its offering rate was 3.50 percent in July 2026, the exact bottom of the target range. Above the range, a standing repurchase facility and the discount window let banks borrow from the Fed if cash gets scarce, which caps how high the overnight rate can spike.

So the Fed does not command the overnight rate. It builds a corridor out of standing offers and lets the market settle inside it. That design matters because it is quiet: the Fed can move the corridor without buying or selling a single bond, which is exactly what the implementation note accompanying every decision spells out.

The Fed does not set the overnight rate. It builds a corridor of standing offers and lets banks settle inside it.

Two goals, or actually three

The phrase you will hear is the dual mandate, meaning stable prices and maximum employment. The statute is slightly different, and the difference is worth ten seconds of your attention.

Section 2A of the Federal Reserve Act instructs the Board and the Committee to promote "the goals of maximum employment, stable prices, and moderate long-term interest rates." That is three goals, not two. The third is usually folded into the first two on the theory that an economy with stable prices will produce moderate long rates on its own, but it is in the law, and it is a reminder that Congress assigned these jobs rather than the Fed choosing them.

Stable prices now has a number attached to it. The Committee defines it as 2 percent inflation over the longer run, measured by personal consumption expenditures prices. Maximum employment has no number, deliberately, because nobody can observe the lowest unemployment rate an economy can sustain without overheating. One goal is a target; the other is a judgment call.

Eight meetings, one statement, and a table of guesses

The Committee meets eight times a year on a published calendar and releases a statement at 2:00 p.m. Eastern on the second day. Four of those meetings also carry a Summary of Economic Projections, the document journalists call the dot plot.

The statement is short, and the useful part is the vote. At the July 29, 2026 meeting the Committee held the range at 3.50 to 3.75 percent by a 9 to 3 vote, with Beth Hammack, Neel Kashkari, and Lorie Logan preferring a quarter point increase. Three dissents is a lot. It tells you the decision was close and that the next one was genuinely uncertain, which no amount of commentary about the chair's body language would have told you.

Then there are the projections, and here the Fed publishes its own error. In the June 2026 projections the median participant expected inflation of 3.6 percent for 2026. Three months earlier the same group's median for the same year was 2.7 percent, and in December 2025 it was 2.4 percent.

Read that sequence carefully. These are not forecasts of a distant decade. They are forecasts of a year that was already underway, made by the people setting the policy, and they moved by more than a full percentage point in nine months. The dot plot is not a promise and it is not a leak. It is a snapshot of what nineteen people currently believe, and their beliefs move.

From one night to thirty years

Here is the chain. The overnight rate anchors what banks will accept for lending money for one night. That sets the price of lending for a week, which sets the price for three months, which anchors Treasury bills, which are the reference point for everything short and safe. Longer rates are built from expectations of where the short rate will average over the life of the loan, plus compensation for the risk of being wrong.

Then the chain reaches you, with lags and with leakage.

Look at what happened between the 2023 average and the first seven months of 2026. The effective federal funds rate fell by 1.40 percentage points. The average rate on credit card balances did not follow: it went from 20.90 percent to 20.97 percent, which is up, not down. The average 60 month new car loan fell about half a point. The 30 year mortgage fell about half a point.

Each of those has its own reason. Credit card pricing carries a large charge for expected losses and for the fact that most of the balance is unsecured, so the policy rate is only one input. Mortgages are priced off the 10 year Treasury rather than the overnight rate, and that long yield answers to inflation expectations more than to the current setting. Car loans sit in between, and depend heavily on what the manufacturer's finance arm is willing to subsidize that quarter.

This is what economists mean by imperfect transmission. It is not a flaw in the plumbing. It is the plumbing.

The balance sheet, and what it is really for

The second tool is the portfolio. When the Fed buys Treasury securities it pays by crediting the seller's bank with new reserves, which is why the exercise is often described as printing money even though no currency is involved. Buying on purpose to push down long rates is quantitative easing; letting the holdings shrink as bonds mature is tightening.

The scale is easy to state and easy to misread. Total assets ran about $0.9 trillion in 2007, reached roughly $9.0 trillion in April 2022, and stood near $6.7 trillion at the end of July 2026, according to the Fed's own balance sheet reporting. The runoff since 2022 was the tightening phase. By mid 2026 the stated aim had shifted to maintaining ample reserves in the banking system, which means the portfolio grows again slowly just to keep pace with a growing economy. Growth in the balance sheet is not automatically stimulus.

While we are clearing up the printing question: the Fed does not print currency. The Board estimates how many notes the country will need and places an order with the Treasury Department's Bureau of Engraving and Printing, which manufactures them and bills the Board for the cost. Physical cash is a Treasury manufacturing operation working to a Federal Reserve purchase order.

What it does not do

Three things get attributed to this institution that belong somewhere else.

• It does not set your mortgage rate. That number tracks the 10 year Treasury yield plus a spread, and both parts move for reasons the Fed only partly influences.
• It does not control prices. It can make borrowing more or less expensive, which changes demand, which eventually changes prices. A tariff, a drought, or an oil disruption moves prices directly and the Fed can only respond after the fact.
• It does not decide fiscal policy. Taxes and spending are Congress, and the Fed's job is to react to their consequences.

Reading the next Fed headline

Three questions will tell you more than any pundit.

First, what changed, the range or the language? A held meeting with a shifted sentence about inflation is often a bigger signal than a cut everyone expected.

Second, how did they vote? Unanimous means settled. Three dissents means the committee is arguing, and arguments predict future moves better than statements do.

Third, is the number you are being shown measured or projected? The target range is a fact with a date on it. A rate path is an assumption wearing a decimal point, and the March to June 2026 revision shows how fast those assumptions move.

Inside Euphoria you can take a live policy decision apart the way a desk analyst would, separating the announced range from the rate that traded, the vote from the statement, and the projection from the measurement, in interactive lessons where getting it wrong costs nothing but a retry.