What compounding actually does. Compounding means your gains start generating their own gains. A dollar earning 7% becomes $1.07 after year one — but the next year's 7% is on $1.07, not $1.00, and so on. The effect is slow at first and then startling: at ~7% real, money roughly doubles every decade. Over a 30-year retirement (plus the decades of saving before it), this exponential curve is why starting early and staying invested matter more than almost any clever tactic.
The long-run numbers. Drawing on the widely cited data assembled by NYU's Aswath Damodaran for 1928–2025, the long-run annual averages look like this: U.S. large-cap stocks ~10% nominal, which after ~3% inflation is roughly 6.5–7% real; 10-year Treasuries ~4.9%; T-bills ~3.3%; inflation ~3%. Stocks pay the most and swing the most; bonds and bills pay less but steady the ride. The "equity risk premium" — the extra return stocks deliver over bonds for bearing their volatility — is the reward that funds most long retirements.
The scale of it. Numbers this abstract are easier to feel with a concrete case: $100 put into U.S. stocks in 1928 would have grown to about $1.16 million by the end of 2025, dividends reinvested. The same $100 in Treasury bills would be a few thousand dollars. That roughly thousand-fold difference over a lifetime is the equity premium compounding across a century — and the reason retirees, who often have 30+ year horizons, generally shouldn't abandon stocks entirely.
Patience beats timing. Any single year can be terrifying — stocks have dropped 30–50% in bad ones. But stretch the window and the picture calms dramatically: across the 1928–2025 record, no rolling 20-year holding period produced a negative total return for U.S. stocks. Even the worst 20-year stretches ground out gains. The practical implication for a retiree with a long horizon: the money you won't spend for 15–20+ years can weather volatility that would be reckless for near-term spending — which is exactly the logic behind the cash/bond buffer for the early years (Chapter 8) and equities for the later ones.
Two honest caveats. First, averages hide sequence risk. A 7% average with the bad years first can still sink a retiree who's withdrawing (Chapter 8). Retirement math cares about the path, not just the destination. Second, the future may not match the past. After a long bull market, U.S. stock valuations (measured by the Shiller CAPE) sit well above historical norms, and bond yields are lower than their century average. Forward-looking models (Morningstar, Vanguard) therefore project returns below the historical averages — which is why their safe-withdrawal numbers (Chapter 7) come in lower than Bengen's backward-looking ones. History is the base rate, not a guarantee.
The practical levers. You can't control returns, but you can control three things that reliably improve outcomes: diversification (spreading across asset classes and geographies so no single bet sinks you), cost (every basis point of fees is a permanent drag on compounding — low-cost index funds keep more of the return), and behavior (staying invested through downturns rather than selling at the bottom, which is where most real-world investors lose to the averages). Compounding does the heavy lifting; your job is mostly to not interrupt it.
The cost of a small drag
Two retirees each start with $1,000,000 earning 7% before fees. One pays 0.1% (index funds); the other pays 1.1% (higher-cost funds). After 25 years the low-cost investor has about $5.0M and the high-cost investor about $3.9M — a ~$1.1M gap created purely by a 1% annual fee compounding. Same market, same returns; costs alone reshaped the outcome.