What Is Compound Interest, and Why Time Does the Heavy Lifting

Compound interest is the mechanism by which money earns money, and then that earned money starts earning too. It is not a trick or a strategy. It is arithmetic, and once the arithmetic is clear the behavior of almost every savings account, retirement plan and credit card balance becomes predictable.

Most explanations stop at "your interest earns interest" and move on. That sentence is true but it teaches nothing about the variables that actually decide the outcome: how often the interest is applied, how the stated rate differs from what you actually receive, why the length of time matters more than the size of the deposit, and why a real investment portfolio never traces the smooth curve a calculator draws.

This guide works through each of those. Every figure below is illustrative -- the returns are assumptions chosen to make the arithmetic visible, not forecasts.

#The formula, one variable at a time

The standard compound growth formula is:

A = P x (1 + r/n)^(n x t)

Four inputs and one output. Each does a specific job.

• P is the principal: the amount present at the start.
• r is the annual interest rate expressed as a decimal, so 6% is 0.06.
• n is the number of compounding periods per year: 1 for annual, 12 for monthly, 365 for daily.
• t is the number of years.
• A is the ending amount, principal and accumulated interest together.

The exponent is where the leverage sits. Changing P scales the answer linearly -- double the principal and you double the result. Changing t moves an exponent, and exponents do not behave politely.

Worked example: Put $10,000 into an account paying an assumed 6% compounded once a year for 10 years. A = 10,000 x (1 + 0.06/1)^(1 x 10) = 10,000 x 1.06^10 = 10,000 x 1.790847 = $17,908.48.

Simple interest on the same deposit would pay 6% of the original $10,000 every year and nothing more: $600 x 10 = $6,000, ending at $16,000. The $1,908.48 gap is the interest that the interest earned. Over 30 years at the same rate the compound figure is $57,434.91 and the simple figure is $28,000 -- the gap has stopped being a rounding difference and become the majority of the return.

#Compounding frequency, and the rate you are actually paid

A "6% account" and a "6% account" can pay different amounts. The stated annual rate -- the nominal rate -- says nothing on its own about how often interest is credited. The effective annual yield does.

Take a nominal 8% and vary only the compounding frequency on a $10,000 deposit for one year.

Two things are worth noticing. Frequency does matter -- daily compounding beats annual by $32.78 on a $10,000 deposit in year one, and that gap itself compounds. But the effect has a ceiling. The jump from annual to monthly captures most of the available benefit; everything beyond daily is a rounding error dressed up as a feature.

This is exactly why US deposit accounts are advertised with an annual percentage yield (APY) rather than a bare interest rate. APY already folds in the compounding frequency, so two APYs are directly comparable while two nominal rates are not. On the borrowing side the equivalent disclosure is APR, which works differently and is worth understanding separately -- see how APR is calculated and what it includes. If you are comparing places to hold cash, how high-yield savings accounts work covers the account mechanics.

Note: A rate quoted "compounded daily, paid monthly" is still compounding daily. What changes monthly is when the credited interest appears on your statement, not when it starts working.

#The rule of 72, and where it stops being reliable

Divide 72 by the annual percentage return and you get an approximate number of years for money to double. At 6%, 72 / 6 = 12 years. At 9%, eight years.

It is an approximation of the exact answer, which is ln(2) / ln(1 + r). Here is how close it gets.

The rule is close enough to use mentally between roughly 4% and 12%, and it drifts at both extremes. At very low rates it overstates the doubling time; at high rates it understates it. For low rates, 70 is a better numerator, which is why demographers use the rule of 70 for population growth.

Its real value is not precision. It is that it turns a percentage into a unit of time, and time is the thing people underestimate.

#Time beats contribution size, and here is the proof

The most repeated claim in personal finance is that starting early matters more than contributing heavily. It is usually asserted rather than demonstrated. Run it properly.

Two people, both assuming a 7% annual return, both contributing $6,000 at the end of each year.

The early starter contributes from age 25 through 34 -- ten contributions, $60,000 total -- then stops completely and never adds another dollar. The balance at 35 is $82,898.69. It then compounds untouched for 30 more years: 82,898.69 x 1.07^30 = 82,898.69 x 7.612255 = $631,045.95 at 65.

The late starter contributes from age 35 through 64 -- thirty contributions, $180,000 total, three times as much money. The balance at 65 is $566,764.72.

The early starter put in a third of the money and finished about $64,000 ahead. The variable doing the work is not discipline or contribution size. It is that the early starter's dollars had 30 to 40 years of exponent behind them while the late starter's had 1 to 30.

This is also the honest answer to the question of whether it is "too late" to start. It is not too late; it is simply more expensive, and the price is measurable.

#Contribution and time combinations

Monthly contributions, 7% assumed annual return compounded monthly, no starting balance.

Compare rows three and five. Both people put in $96,000. The one who spread it over 40 years at $200 a month ends with $524,963; the one who compressed it into 20 years at $400 a month ends with $208,371. Same money in, two and a half times the difference out.

Notice also that growth only overtakes contributions somewhere between year 15 and year 20 at these assumptions. The first decade feels like nothing is happening. That is not a failure of the plan; it is the shape of the curve. Understanding this is most of what stops people abandoning a plan in year four.

The calculator below runs this arithmetic for any combination of starting balance, monthly contribution, assumed rate and time you enter.

#A year-by-year view

Averages hide the mechanism. This is $10,000 starting balance plus $3,000 added at the end of each year, at an assumed 7% annual return.

Total contributed across ten years: $40,000. Total interest: $21,120.86. In year one the interest was less than a quarter of the contribution. By year ten it exceeded it. Nothing changed except the size of the base the rate was applied to.

#Compounding in reverse: debt

The same formula runs against you when you owe money. A credit card charging an assumed 22% APR compounds your balance monthly at 22% / 12 = 1.8333% per month.

Worked example: A $5,000 balance at an assumed 22% APR. Month one interest is 5,000 x 0.018333 = $91.67. If the minimum payment is 1% of the balance plus interest, subject to a $25 minimum-payment floor, that month's payment is $141.67, of which $50 reduces principal. The floor only starts to bite once the balance falls below about $880, because 1% plus interest is 2.833% of the balance and 2.833% of $880 is roughly $25. Continue that pattern and the balance takes about 230 months -- roughly 19 years -- and costs about $8,100 in interest, more than the original balance.

Change one variable. Pay a fixed $150 a month instead of the shrinking minimum and the same $5,000 clears in 52 months with about $2,796 in interest. Pay $250 and it clears in 26 months with about $1,071.

The reason minimum payments behave so badly is structural. They are calculated as a percentage of the outstanding balance, so as the balance falls the payment falls with it, and the schedule stretches out indefinitely. A fixed payment does the opposite: as interest shrinks, more of the same dollar amount goes to principal. Methods for attacking multiple balances are covered in practical approaches to clearing card balances.

Warning: A guaranteed 22% saved by clearing high-rate debt is arithmetically better than an uncertain 7% earned in a market. When the two compete for the same dollar, the debt usually wins on the numbers alone.

#Inflation: nominal growth versus real growth

A 7% return in an economy with 3% inflation is not a 7% increase in what you can buy. The correct adjustment is not subtraction but division:

Real return = (1 + nominal) / (1 + inflation) - 1

(1.07 / 1.03) - 1 = 0.03883, or 3.883%. Subtracting gives 4%, which is close enough for mental arithmetic at low rates and drifts badly at high ones.

The same adjustment applies to the ending balance. If prices rise at an assumed 3% a year, $100,000 in 25 years buys what $100,000 / 1.03^25 = $47,760 buys today. A projection that looks comfortable in nominal dollars can be roughly half as comfortable in real ones.

This is the single most common omission in retirement projections. Two fixes work: run the projection using a real rate of return and read the answer as today's dollars, or run it nominally and then deflate the answer. Doing neither, or doing both, produces numbers that mean nothing. Recent inflation data is published by the Bureau of Labor Statistics.

#What the usual advice gets wrong

The smooth curve is a lie of convenience. A calculator applies exactly 7% every year. A diversified stock portfolio does no such thing. It might return 22%, then -9%, then 14%, then -3%. The arithmetic average of those four is 6%; the actual compounded result is lower.

Here is why. A sequence of +20% then -10% has an arithmetic mean of 5%, but $10,000 becomes $10,000 x 1.20 x 0.90 = $10,800, a compound rate of 3.92% per year. A steady +5% and +5% produces $11,025. Same average, different outcome. Volatility drags the compound result below the arithmetic mean, always, and the drag grows with the size of the swings.

Losses are asymmetric. A 50% loss requires a 100% gain to recover, not a 50% gain. A 20% loss requires 25%. This is why controlling downside matters disproportionately, and why how you allocate across asset classes is a compounding decision rather than merely a risk-tolerance one.

Sequence of returns matters at the ends. While you are accumulating and adding money, an early downturn is arguably helpful -- your contributions buy more shares cheaply. Once you are withdrawing, an early downturn is damaging, because you sell assets at depressed prices and the remaining base is permanently smaller. Two retirees with identical average returns over 30 years can end up in very different positions purely because of the order in which those returns arrived.

Fees compound too. An expense ratio is subtracted from the return before compounding, every year, whether the market rises or falls. Over 30 years a 0.60% annual cost instead of 0.03% on $100,000 growing at an assumed 7% is the difference between roughly $754,800 and roughly $643,100. The mechanics are unpacked in how expense ratios and investment fees work.

Taxes interrupt compounding in a taxable account. Interest taxed each year reduces the base that next year's return applies to. Tax-advantaged accounts such as an IRA or a workplace plan let the full amount compound and settle the tax at a single point -- one reason the Roth versus traditional decision is worth the effort.

#Making the arithmetic work in practice

Four levers exist and only four: how much you start with, how much you add, what rate you earn, and how long you leave it. You control the first two completely, the third only indirectly through cost and allocation, and the fourth only by starting now rather than later.

Contributing on a fixed schedule rather than in occasional lumps also removes the decision of when to invest, which is a decision most people make badly -- dollar-cost averaging covers the trade-offs. And none of this works if a surprise expense forces you to liquidate at a bad moment, which is the practical argument for holding a cash buffer first.

Compounding is not fast. It is relentless, which is a different property, and a considerably more useful one.