Last updated September 19, 2026

Methodology

This page lists every source and series behind the figures on Inflation Money, the formulas that turn them into clocks and calculators, and the limitations readers need to know about.

Sources

A scheduled job retrieves each series below, regenerates JSON snapshots and commits them to the site. Every figure carries the date of the observation it is anchored to. Monthly BLS series are seasonally adjusted unless the series ID says otherwise.

Formulas

Purchasing-power loss. The value today of an amount from an earlier date is the amount multiplied by the earlier CPI-U index level and divided by the latest level. The percentage loss is one minus the ratio of the earlier index to the latest index.

Inflation rates. Headline and core inflation are the percentage change from the observation twelve months earlier. The inflation loss calculator's 3, 5 and 10-year rates are annualised: the ratio of the latest index to the index that many years earlier, raised to the power of one over the number of years, minus one.

Live inflation clock. The clock at /inflation-clock/ compounds a published annual rate down to shorter intervals rather than dividing it. A fixed amount A has the buying power of A ÷ (1 + r)^t after t years at annual rate r, so the buying power lost over an interval of length t is A × (1 − (1 + r)^−t), with a year taken as 365.25 days. Compounding is what makes the intervals consistent with each other: the per-second figure repeated 31,557,600 times reproduces the per-year figure, whereas an even split of the annual rate would overstate the yearly loss (at 3.3%, $10,000 loses $319.46 a year, not $330). The short-window bases annualise a seasonally adjusted index over 1, 3 or 6 months as the ratio of the latest index to the index that many months earlier, raised to the power of twelve over the number of months, minus one. The main readout starts at the amount entered and falls from the moment the clock starts, which is page load, a reset, or a change of amount; the rate it falls at is extrapolated to the present from the most recent published observation, so the underlying data is monthly while the decay is continuous. A separate figure reports what the same amount has lost since that observation was published, and that one does not reset. Negative rates are carried through unchanged, so a deflationary rate counts buying power gained.

Annual pace. For the national debt, the yearly pace is the change between the latest Treasury row and the first row at least 330 days earlier, divided by the elapsed time in years. A window of roughly a year keeps a single large auction or redemption from distorting the clock.

Projection between refreshes. A ticking clock shows the last observed value plus the yearly pace multiplied by the fraction of a year elapsed since the observation date. The projection is linear and re-anchors whenever a new snapshot is published. Point-in-time series, including CPI index levels, M2, GDP, federal receipts and outlays and the gold and Bitcoin comparisons, show their latest observed value and do not tick.

Per-person figures. Debt per citizen divides the national debt by the latest resident population from FRED series POPTHM. Debt per household divides by an estimated 132 million households, and debt per taxpayer by an estimated 169 million individual income tax filers. Both bases are fixed estimates and both figures are labelled as estimates wherever they appear.

Budget balance. The federal deficit shown on the site is federal current expenditures minus federal current receipts, so the two sides of the calculation come from the same BEA source family.

Market comparison. Each selected asset is rebased to $10,000 on the first month shared by all selected series and tracked forward monthly. The inflation-adjusted view divides each value by the ratio of that month's CPI-U to the starting month's CPI-U. Three-month Treasury bill yields are converted into a monthly compounded return so cash is comparable.

Known limitations

• Source agencies revise past data. A revision reaches the site the next time the snapshot is regenerated.
• The projection assumes the recent annual pace continues unchanged until the next release, so a ticking clock drifts from the true figure as time passes since the last observation.
• The household and taxpayer bases used for per-household and per-taxpayer figures are fixed estimates, not live series.
• Gold, silver and the market comparison series use monthly averages, so they differ from any single day's spot price or close.
• The Bitcoin price is the daily Coinbase price published by FRED, so it reflects one exchange rather than a cross-exchange average.
• If the scheduled refresh fails, the site keeps showing the previous snapshot. The "as of" date on each metric shows how old the observation is.
• CPI-U measures a national average basket; individual households experience different inflation rates.
• Displayed values are rounded; calculations use the unrounded snapshot values.

Questions about a source or formula can be sent to [email protected].