Which Number to Plan With

Inflation has not always been this calm — it peaked above 14% in 1980 ( Figure 2.1), and the long-run CPI-U average since the series began in 1913 runs right around 3.2% annually.

That century figure is quoted everywhere, including in earlier editions of this book, and it is the wrong number to plan with. It is a geometric mean taken across three monetary regimes, two of which no longer exist: a gold-convertible dollar, then Bretton Woods, then a fiat dollar with no inflation target at all until the 1990s. It is also dominated by a single episode. Roughly a fifth of the entire century’s price growth occurred in the ten years from 1970 to 1980 — nine percent of the elapsed time. Remove 1970–1983 and the remaining ninety-nine years compound at 2.6%.

Every window drawn from the current regime lands lower and lands close together:

Two forward-looking anchors agree with the bottom of that range. The Federal Reserve targets 2% on the PCE index, and CPI-U runs structurally about 0.4 points above PCE because of formula, weighting, and scope differences — so a central bank hitting its target implies roughly 2.4% CPI-U. The bond market’s breakeven rate, the spread between nominal Treasuries and TIPS of matched maturity, has sat near the same level; because TIPS principal is indexed to CPI-U itself, that spread is directly comparable with no adjustment. Three independent methods — realized experience under the current regime, the policy target, and the traded price of inflation — converge on the same neighborhood. That convergence, not any one of them alone, is the justification.

This book uses two deflators, not one. A single rate applied everywhere is wrong in both directions at once, because the cost of the error flips sign depending on what you are deflating.

2.5% –- for real returns, discount rates, and hurdles.

Use this whenever you are asking whether an investment beat inflation, computing a real rate of return, or discounting a future cash flow. Overstating inflation here manufactures a hurdle that does not exist and pushes you into more risk than you need to carry.

3.5% –- for spending and retirement-need projections.

Use this when you are sizing a future liability: the retirement number, the college number, the survival floor. Two reasons it is higher. The error is asymmetric — underfunding a thirty-year retirement is not recoverable, while overfunding it is — and the basket this book’s reader actually buys is not the headline basket. Private tuition, healthcare, home services, insurance, and property tax have run persistently above CPI-U.

There is a third case worth naming because it is where the mistake is most expensive. Decisions about fixed-rate debt should use the lowest rate you can defend, which today means the traded breakeven. A high assumed inflation rate makes a fixed mortgage look like free money and argues against ever paying it down, and that argument is only as good as the inflation assumption underneath it (section “Using Debt Well: A Decision Framework”).

Better than any default is your own number. Compute a weighted rate across your own spending categories; the method is in section “Measuring and Forecasting Inflation”, along with which published index governs which of your obligations — because “inflation” is not one number, and the differences between CPI-U, CPI-W, chained CPI, and PCE decide your tax brackets, your Social Security raise, and your TIPS principal, respectively.

What matters for the rest of this chapter is only the arithmetic: whatever rate you choose, it compounds, and it compounds against you.