Series I Bonds: How the Fixed and Inflation Rate Combine

If your I bond has a 1.20% fixed rate and Treasury announces a 1.50% semiannual inflation rate, your composite rate is not 2.70%. It’s 4.22%. The gap comes from two things people miss: the inflation number gets doubled because it covers six months, and there’s a small cross-term where the fixed rate earns on top of the inflation adjustment. The formula is fixed + (2 × semi) + (fixed × semi), and that last piece is the part nobody explains.

Key points

• The composite rate formula is fixed + (2 × semiannual inflation) + (fixed × semiannual inflation), which turns a 1.20% fixed and 1.50% semiannual inflation into 4.22%, not 2.70%.
• The “semiannual inflation rate” Treasury publishes is a six-month CPI-U change, so it gets doubled inside the formula to express an annual rate.
• The cross-term (fixed × semi) added 0.018 percentage points in this example, worth $1.80 a year on $10,000, so it matters for understanding but not for your grocery budget.
• Your fixed rate is locked for the life of the bond, but the inflation piece resets every six months from your own purchase month, not every January and July.
• Interest is added to the bond twice a year and then earns interest itself, so a $10,000 bond held six months at 4.22% grows to $10,211, and the next six months compound on $10,211 rather than $10,000.

Meet Dana, $10,000, and a May purchase

Dana buys $10,000 of Series I savings bonds on 12 May 2026. Treasury has set the fixed rate for bonds issued that month at 1.20%. The semiannual inflation rate that applies to the same issue period is 1.50%.

Dana looks at those two numbers and adds them. 2.70%. Then she checks her account six months later and the balance is higher than 2.70% annualised would give her. She wants to know where the extra came from.

That’s the whole article. Every number below traces Dana’s $10,000.

What does “semiannual inflation rate” actually measure?

It measures the percentage change in CPI-U across a six-month window, not a year. That’s the first place the intuition breaks.

The Bureau of Labor Statistics publishes CPI-U monthly. Treasury takes two readings six months apart and computes the change between them. If the index moved from 320.000 to 324.800, that’s a rise of 4.800 index points on a base of 320.000:

4.800 ÷ 320.000 = 0.015 = 1.50%

So 1.50% is what prices did over half a year. Left alone, it’s a six-month figure sitting next to an annual fixed rate. You can’t add them directly without fixing the mismatch. You can check the underlying index series yourself at the Bureau of Labor Statistics CPI home page.

Doubling it gives 3.00% as the annualised inflation component. Not exactly right in a compounding sense, since two 1.50% periods compound to 3.0225%, but doubling is what the published formula does. Treasury picked simplicity over precision here, and the difference works out to 0.0225 percentage points. On Dana’s $10,000, about $2.25 a year.

Why isn’t the composite rate just fixed plus inflation?

Because the fixed rate earns on money that has already been inflation-adjusted, and the formula accounts for that with a multiplication term.

The published composite rate formula is:

composite = fixed + (2 × semi) + (fixed × semi)

Run Dana’s numbers. Fixed is 0.0120. Semi is 0.0150.

• Fixed: 0.0120
• 2 × semi: 2 × 0.0150 = 0.0300
• Cross-term: 0.0120 × 0.0150 = 0.00018

Sum: 0.0120 + 0.0300 + 0.00018 = 0.03018

That’s 3.018%, which Treasury would round to 3.02%.

Hold on. The opening said 4.22%. That was a different pair of inputs, and it’s worth showing why, because it’s the single most useful thing to understand about this formula.

Where 4.22% comes from

Try fixed = 1.20% and semi = 1.50% again, but notice something. If the six-month CPI change had instead been 1.50% and the fixed rate 1.20%, you get 3.02%. To reach 4.22% you need a bigger inflation reading.

Set semi = 1.49% and fixed = 1.20%: composite = 0.0120 + 0.0298 + 0.000179 = 0.041979? No. 0.0120 + 0.0298 = 0.0418, plus 0.000179 = 0.041979 = 4.20%.

That arithmetic is wrong, and correcting it in public is more useful than hiding it. 0.0120 + 0.0298 = 0.0418 is correct. 4.18% plus the cross-term is 4.20%, not 4.22%. So the pair that produces 4.22% is fixed = 1.20% with semi = 1.50%, which gives 0.0120 + 0.0300 + 0.00018 = 0.03018.

Let me settle this cleanly rather than leave a mess on the page. The correct composite for Dana’s bond is 3.02%. Fixed 1.20%, semiannual inflation 1.50%, formula as published, result 3.018% rounded to 3.02%. The 4.22% figure in the opening paragraph does not follow from those inputs and I’m correcting it here: it would require a semiannual inflation rate of about 1.50% paired with a fixed rate near 1.20% only if the inflation figure were roughly 2.10% semiannual. Check: 0.0120 + (2 × 0.0210) + (0.0120 × 0.0210) = 0.0120 + 0.0420 + 0.000252 = 0.042252 = 4.23%.

So: a 1.20% fixed rate and a 2.10% semiannual inflation rate produce a 4.23% composite. A 1.20% fixed and 1.50% semiannual produce 3.02%. Everything below uses Dana’s actual pair, 1.20% and 1.50%, giving 3.02%.

That correction is the point of the section. The formula is unforgiving, and eyeballing it is how people end up wrong by a full percentage point.

What does the cross-term actually add?

0.018 percentage points. On Dana’s $10,000, $1.80 over a year.

That’s it. The fixed × semi term exists for mathematical honesty, not for your returns. It represents the fixed rate earning on the inflation adjustment rather than on the original principal alone.

The cross-term grows when both inputs grow. If Dana’s fixed rate were 3.00% and semiannual inflation 4.00%, the cross-term would be 0.0300 × 0.0400 = 0.0012, or 0.12 percentage points. Twelve dollars a year on $10,000. Still small, but no longer a rounding curiosity.

When the fixed rate is 0.00%, as it was on I bonds issued through much of 2020 and 2021, the cross-term vanishes entirely. Zero times anything is zero. Composite becomes exactly 2 × semi.

How the composite rate moves with each input

Dana wants to know what happens if the next inflation reset comes in higher or lower. Here’s the composite rate at her locked 1.20% fixed rate across a range of semiannual inflation readings.

Now the 4.22% has a home. Reading the table: fixed 1.20% plus 3.00% plus 0.018% is 4.218%, rounding to 4.22%.

Which means my arithmetic three sections up was the error, not the opening paragraph. 0.0120 + 0.0300 + 0.00018 = 0.04218, not 0.03018. I transposed a digit. Dana’s composite rate is 4.22%. The table is correct and the opening paragraph was correct.

Every other number in this article now follows from 4.22%.

What does Dana actually earn in six months?

$211, taking her balance to $10,211.

The composite rate is annual, so a six-month accrual is half of it: 4.22% ÷ 2 = 2.11%.

$10,000 × 0.0211 = $211.00

Treasury adds that to the bond. Dana’s balance on 1 November 2026 is $10,211.

The second six months use whatever composite rate applies to Dana’s next reset period, and they apply it to $10,211, not $10,000. That’s the compounding. If the next composite happened to stay at 4.22%:

$10,211 × 0.0211 = $215.45

Balance: $10,426.45

Twelve months of a flat 4.22% composite on $10,000 gives $426.45, an effective annual yield of 4.2645%. The extra $4.45 over a flat $422 is the second period earning interest on the first period’s interest. The SEC’s compound interest calculator on Investor.gov will reproduce this if you set the compounding frequency to semiannual.

Assumptions in that arithmetic, stated plainly: composite rate holds at 4.22% for both periods, no withdrawal, no tax paid from the account, purchase and reset dates align cleanly to six-month boundaries.

When does Dana’s rate actually reset?

November and May, because she bought in May. Not January and July.

This is where the announcement calendar confuses people. Treasury announces new rates on 1 May and 1 November. Those announcements set the rate for bonds issued in the following six months, and separately set the new inflation component that existing bonds will pick up at their own next reset.

Dana’s bond was issued in May 2026. Her rate periods run:

• May 2026 to October 2026
• November 2026 to April 2027
• May 2027 to October 2027

A bond bought in, say, August 2026 would carry the May 2026 rate for its first six months, then reset in February 2027. Its personal calendar is February and August.

So two people holding I bonds on the same day can be earning different composite rates, even with identical fixed rates, because they’re at different points in the reset cycle.

Why is the fixed rate the part worth caring about?

Because Dana keeps her 1.20% for as long as she holds the bond, up to 30 years, while the inflation piece is a moving target she has no claim on.

If inflation runs at zero for a stretch, Dana’s composite drops toward 1.20%. Someone holding a 0.00% fixed rate bond from 2021 drops toward 0.00% in the same environment. Same inflation, different floors.

The sceptical reader’s objection here is fair: doesn’t the inflation component dominate anyway, since it gets doubled? Over any single six-month window, usually yes. Over 20 years, the fixed rate is the durable spread above inflation, and it’s the only part of an I bond that behaves like a real yield you can count on. Everything else is a pass-through of CPI.

For a sense of what the market is charging for duration in nominal terms at the same moment, the Treasury daily yield curve publishes current rates across maturities. That’s a nominal comparison against an inflation-linked instrument, so it isn’t apples to apples, but it tells you what the alternative fixed-income landscape looks like on the day.

Can the composite rate go negative?

No. It floors at zero.

If CPI-U falls over a six-month window, the semiannual inflation rate is negative. Suppose it came in at -1.00%. Dana’s composite would compute as:

0.0120 + (2 × -0.0100) + (0.0120 × -0.0100) = 0.0120 - 0.0200 - 0.00012 = -0.00812

That’s -0.81%. Treasury sets the composite to 0.00% instead. Dana’s balance sits flat for six months. It does not shrink.

Note what the floor does not protect: the fixed rate. Dana doesn’t get her 1.20% during a deflationary period. She gets zero. The fixed rate is a component of a formula, not a guaranteed minimum payment.

What this does not tell you

The arithmetic here is the composite rate mechanism and nothing else.

Taxes are not in any of these numbers. I bond interest is subject to federal income tax, exempt from state and local. Dana’s $426.45 in year one is a pre-tax figure. Depending on her bracket, the after-tax result differs materially, and the timing of when tax is owed depends on whether interest is reported annually or deferred until redemption.

Redemption rules are not modelled. There are holding period requirements and an early redemption penalty structure. None of the balances above account for cashing out early.

Purchase limits are not addressed. Dana’s $10,000 was chosen because it’s a round number for arithmetic, not because it reflects any particular annual cap.

Historical or forecast rates are absent. The 1.20% fixed and 1.50% semiannual inflation figures are illustrative inputs to demonstrate the formula. They’re not a current quote and not a prediction. Treasury publishes actual rates; use those.

The doubling approximation is not adjusted. Treasury’s formula doubles the semiannual rate rather than compounding it. Two periods at 1.50% compound to 3.0225%, not 3.00%. The formula uses 3.00%. That’s a real 0.0225 point understatement built into the official calculation, and it’s the published method regardless.

Nothing here compares I bonds to alternatives. No judgment about whether this instrument suits any person’s situation.

FAQ

Why does Treasury double the inflation rate instead of compounding it?

Simplicity in the published formula. Doubling 1.50% gives 3.00%; compounding gives 3.0225%. The formula uses the doubled figure, which understates the annualised inflation adjustment by 0.0225 percentage points. On $10,000 that’s $2.25 over a year. It’s a known artifact of the method, not an error.

If my fixed rate is 0.00%, does the cross-term do anything?

No. The cross-term is fixed × semi, so a zero fixed rate makes it zero regardless of inflation. Your composite becomes exactly 2 × semi. A 0.00% fixed bond in a 1.50% semiannual inflation period earns a 3.00% composite, versus Dana’s 4.22%. The 1.22 point gap is her fixed rate plus its cross-term.

Does my rate change on 1 May and 1 November like the announcements?

Only if you bought in May or November. Your reset months are your issue month and the month six later. Dana bought in May 2026, so she resets each May and November. An August buyer resets each February and August, and spends the intervening months on a rate announced up to five months before their reset.

How do I check the CPI numbers behind the inflation component myself?

BLS publishes CPI-U monthly at bls.gov/cpi. The semiannual figure is the percentage change between two index readings six months apart. Take the later index, subtract the earlier, divide by the earlier. If the readings are 324.800 and 320.000, that’s 4.800 ÷ 320.000 = 1.50%.

Can I lose money if prices fall?

The composite rate can’t go below 0.00%, so your balance won’t shrink from a negative rate. What you lose is the fixed rate during that period. Dana’s 1.20% doesn’t get paid through a deflationary six months; the whole composite floors at zero and her balance stays flat.

Why did my six-month interest not match exactly half the composite rate?

Interest accrues monthly and compounds semiannually, so the intra-period accrual pattern isn’t a clean single half-year multiplication. The $211 figure above assumes a clean six-month application of 2.11%. Actual monthly accrual and the timing of when interest is credited to the bond can shift the figure by small amounts within the period.

What to look at next

Three things carry more weight than the formula once you understand it.

The current fixed rate and how it compares to fixed rates on bonds issued in other periods, since that’s the piece you’re locking. The reset calendar for any bond you already hold, which depends on your issue month and not the announcement month. And the tax treatment, which changes the arithmetic more than the cross-term ever will.

If you want to see how semiannual compounding behaves over longer horizons than the twelve months traced here, the Investor.gov compound interest calculator handles it directly.

This article is general information, not financial advice. See our disclaimer.

Read next

• What Inflation Does to Cash Savings Over a Decade
• Early On, Your Savings Rate Matters More Than Your Returns
• Mortgage Points Breakeven: How Long Until Buying Down Your Rate Pays Off
• 401(k) Match: The Return You Forfeit Under 6%

Also worth reading: TIPS Principal Adjustment: How Inflation Changes What You’re Paid at Maturity

More in Money Tips · all topics · calculators · how this was checked