A few years ago, a colleague of mine pulled up the TSP's performance page, sorted by the "since inception" column, and announced she was moving everything into the C Fund. "Ten percent a year," she said. "Why would I own anything else?" She was three years from retirement.

She wasn't wrong about the number. She was wrong about what the number could do for her.

Historical compound annual growth rate (CAGR) is the most quoted statistic in retirement planning, and the least understood. Let's look at what it shows for each TSP fund, and then at the parts of the story it leaves out.

What the numbers roughly look like

CAGR is the single steady yearly rate that would take you from your starting balance to your ending balance. Real returns never arrive that smoothly, but the figure makes funds easy to compare.

Here are approximate long-run figures. Exact numbers change every month, so check tsp.gov for current performance before you rely on any of them.

  • C Fund (large U.S. stocks): roughly 10% annualized since 1988
  • S Fund (small and mid-size U.S. stocks): roughly 9% since its 2001 start
  • I Fund (international developed stocks): roughly 4% to 5% since 2001
  • F Fund (U.S. bonds): roughly 4% to 5% since 1988
  • G Fund (special Treasury securities): roughly 4% to 5% since 1988

Notice the awkward part already. The C, F, and G funds start in 1988. The S and I funds start in 2001. You are comparing different time periods, and that matters more than it sounds.

Problem one: the start date does the heavy lifting

The I Fund's modest number partly reflects that it launched right as international stocks entered a long stretch of lagging U.S. markets. Run the clock from a different year and the ranking can shuffle.

The C Fund shows this clearly. From 2000 through 2009, it delivered a slightly negative annualized return. A full decade, essentially nothing. Yet the "since inception" figure sits near 10% because the 1990s and the 2010s were so strong on either side.

So when you see a CAGR, ask two questions. When does the clock start, and when does it stop? A number measured from a market low looks heroic. A number measured from a peak looks grim. Neither is lying, and neither is predicting anything.

Problem two: CAGR smooths over the ride

Two funds can share a CAGR and feel completely different to own.

Consider the gap between average return and compound return. Say your account gains 50% one year and loses 50% the next. The simple average is 0%. But $10,000 becomes $15,000, then $7,500. You are down 25%, and your CAGR is negative.

Real drawdowns are the same story at a larger scale. In 2008, the C Fund lost about 37%. The I Fund lost more than 40%. The G Fund, meanwhile, never posts a negative calendar year. In 2022, the C Fund fell around 18% and the F Fund fell about 13%, which surprised plenty of people who thought bonds were the safe seat.

A 10% CAGR is the finish line view. Nobody lives there. You live in the individual years, and your nerves, your job situation, and your withdrawal plans all operate in real time.

Problem three: your money did not arrive in a lump sum

CAGR assumes one deposit at the beginning and no activity afterward. That is not how a TSP account works.

You contribute every pay period. Some of your money has compounded for 25 years. Some has been in for six months. Your personal return depends on when your dollars went in, which is why two coworkers in the same fund can have different results.

Contributions made right before a downturn buy shares at high prices. Contributions made during a downturn buy them cheap. Over a full career, regular investing tends to smooth out some of the timing risk, but it does not eliminate it, and the fund's published CAGR will not reflect your experience.

Problem four: withdrawals change everything

This is the one that bites near retirement.

Imagine a retiree with $800,000 who plans to withdraw $32,000 a year, a 4% rate. If she is fully in stocks and a 2008-style 37% decline hits in year one, her balance drops to about $504,000. Her next $32,000 withdrawal is now more than 6% of what is left. She is selling shares at low prices to cover groceries, and those shares are not around to recover later.

Now picture the same decline ten years into retirement, after the account has grown. The long-run CAGR could end up identical in both cases, yet the outcomes differ sharply. Planners call this sequence of returns risk. A CAGR tells you nothing about sequence, because it only knows the beginning and the end.

This is the real reason my colleague's plan worried me. At 10%, a stock-heavy portfolio looks wonderful on a spreadsheet. With three years to go and a withdrawal plan ahead, the order of returns matters more than the average.

Problem five: the G Fund's number is not a promise

The G Fund's historical figure comes from decades in which interest rates were far higher than they have been recently. Its rate is tied to Treasury yields, so it moves with the rate environment. The 1990s G Fund and today's G Fund are different animals.

Likewise, a historical F Fund CAGR reflects a long bond bull market as rates fell from their early 1980s peaks. Falling rates lifted bond prices for decades. Starting yields are a better guide to future bond returns than the past CAGR is.

Inflation and taxes sit outside the number

Published fund returns are nominal. A 4.5% return during a 3% inflation year leaves you with about 1.5% in real purchasing power. The G Fund has had stretches where it barely kept pace.

Taxes also change the picture. Traditional TSP withdrawals are taxed as ordinary income, while qualified Roth withdrawals are not. Two accounts with identical CAGRs can leave you with different spendable amounts. Check irs.gov for current rules on contribution limits and distributions, since those figures shift yearly.

So how should you use CAGR?

Treat it as a rough description of how asset classes have behaved, not a forecast.

A few habits help:

  1. Compare funds over the same period. Do not put a 1988 start next to a 2001 start and call it a ranking.
  2. Look at the worst years, not just the average. The drawdown tells you what you will have to sit through.
  3. Match your allocation to your timeline. Money you need in five years has a different job than money you need in twenty-five.
  4. Consider your withdrawal plan. A portion in the G Fund can act as a buffer so you are not forced to sell stocks in a slump.
  5. Review lifecycle funds as an alternative. L Funds adjust their mix over time, which addresses the timing problem for people who prefer not to manage it.

My colleague did not end up all in the C Fund. We looked at the worst year, ran the withdrawal math, and she settled on a mix she could sleep with. She gave up a little of the headline return and bought a lot of certainty.

That trade is worth considering. The number on the performance page is real, but it is a summary of the past, and you will be investing through the future one year at a time.