Explainer · interactive · economics

How a higher interest rate lowers inflation

A central bank raises the one rate it sets. Loan and savings rates follow, people and firms borrow and spend less, and with less demand, firms raise prices and wages more slowly. Inflation falls last, typically one to two years later.

That is the mainstream view, and its direction is well supported. How large the effect is, how long it takes, and two claims about whether a hike can raise inflation are disputed. Below, a small constructed model lets you switch each assumption and see what it does.

Level 1 · what happens

The chain

policy rate ↑→ market rates ↑→ real rate ↑→ demand ↓→ output gap < 0→ inflation ↓ The policy rate is the rate the central bank sets on overnight money between banks. The real rate is an interest rate minus the inflation people expect: it is what borrowing really costs, and it rises because expected inflation does not jump with the hike. The output gap is how far the economy produces below (−) or above (+) what it can sustain without speeding up inflation, in %. And if people believe the bank, the inflation they expect falls too, which lowers inflation directly.
Predict first

When does inflation respond?

In the model below, the bank raises its rate by 1 point, holds it for 4 quarters (one year), then lowers it again. Expectations are anchored: people expect inflation near the target.

The instrument · a constructed toy model

Switch the assumptions

Each line is a change from a steady state with inflation at target, in percentage points, over 16 quarters. An output gap of −0.63% means the economy produces 0.63% below what it can sustain. The numbers show directions and order, not sizes for any real economy. The model leaves out the expectations channel: here, expected inflation never responds to the hike itself, only to past inflation.

Quarters after the hike begins. Hover for values.

Assumptions

Output gap, lowest
Inflation, lowest
Inflation, highest
Inflation, quarter 16
Go deeper

Channels, reasons, and the model

Level 2 · how

The channels
  1. Borrowing. Variable-rate loans cost more at once; new fixed-rate mortgages cost more as they are taken out or renewed. Households and firms borrow and spend less.
  2. Saving and assets. Saving pays more; bond and house prices fall, so people feel less wealthy and spend less.
  3. The currency. Higher rates attract money from abroad, the currency rises, imports get cheaper, and exports harder to sell.
  4. Expectations. If people believe the bank will bring inflation down, they ask for smaller wage and price rises now.
Level 3 · why

Why the real rate, why a lag, why expectations

The real rate. A loan at 5% while prices rise 5% a year costs nothing in goods. What changes spending is the real rate. If expected inflation rose one for one with the policy rate, borrowing would cost no more and demand would not fall. The neo-Fisherian toggle shows that case: inflation ends 1 point higher.

The lag. In real economies many loans are fixed for years, prices are set in contracts and menus, and wages are agreed once a year, so spending turns first and prices last, typically one to two years after a change. The toy copies only the order: output turns when the hike ends, at quarter 5, and inflation two quarters later, at quarter 7.

Expectations. Wages and prices are set for the months ahead, so they carry the inflation people expect. When expectations are anchored, a supply shock fades: in the toy model, a 2-point energy shock is back to 0 by quarter 16 with no hike. When they follow last year's inflation, the same shock peaks at +2.40 and is still +0.44 at quarter 16. Weak anchoring also deepens a hike's effect: as inflation falls, expected inflation falls, the real rate rises further, and demand falls more, so inflation drops −0.41 points instead of −0.24, and stays lower longer. That is why central banks guard their credibility.

Level 4 · the model

The toy model's equations

Quarterly, all in changes from a steady state. Δi is the change in the policy rate, y the output gap, π the change in inflation, πe expected inflation, and w how strongly expectations are anchored to the target. The coefficients are chosen to show the order of effects, not estimated. Market rates move fully and at once with the policy rate, and expectations look only back.

output gap          y(t)  = 0.7 · y(t−1) − 0.25 · (Δi(t−1) − πe(t−1))
expected inflation  πe(t) = (1 − w) · π(t−1)     w = 0.9 anchored, 0.4 weakly anchored
inflation           π(t)  = πe(t) + 0.35 · y(t−2) + [cost] 0.15 · Δi(t−1) + [shock]
neo-Fisherian       πe(t) = Δi(t), and demand does not respond

The energy shock adds 2 points in quarter 1 and shrinks by 40% each quarter. Real models used by central banks have many more equations and estimated coefficients; their lags for inflation are one to two years.

What to trust

Established, estimated, disputed

establishedThe direction.

The policy rate moves market rates; a higher real rate lowers demand; lower demand slows price and wage rises.

estimateThe size and the lag.

Central banks put the largest effect on inflation one to two years out. Studies differ by country and period.

disputedThe price puzzle.

Some studies find prices rise for a while after a hike. Firms passing on higher financing costs could explain it; so could studies that miss what the central bank knew when it moved.

disputed · minorityThe neo-Fisherian view.

A permanently higher rate raises inflation in the long run, because expected inflation adjusts to it. Most central banks do not act on this view.

constructedEvery number from the model.

A toy with chosen coefficients. It shows what each assumption does, not how much for any economy.

In short

Summary

  1. A higher policy rate lowers inflation by raising the real cost of borrowing, which lowers demand.
  2. Spending turns first; inflation responds last, typically one to two years later.
  3. Anchored expectations make the job easier: shocks fade instead of feeding on themselves.
  4. Two disputed views say a hike can raise inflation: briefly (the cost channel), or for good (neo-Fisherian).

Not covered: quantitative easing; fiscal policy; how the neutral rate is estimated; exchange-rate regimes.