When prices rise and the economy feels uncertain, how do you preserve your purchasing power? Stocks? Bonds? Precious metals? Let's explore gold—and what its history tells us about protecting buying power over time. Below, choose an amount, purchase date, and comparison date to see how gold held up against everyday costs, then explore how different holding periods and monthly purchases changed the historical results.
The calculator and chart need JavaScript and the saved data file.
Purchasing-power change
- Gold acquired
- Ending gold value
- Needed to keep up with consumer prices
- Above / below that amount
Gold value versus the inflation benchmark
- Gold value
- Amount needed to keep up with consumer prices
But what if you bought at the wrong time?
January 1980–January 2000
- Value of a $100 gold purchase
- Amount needed to keep up with overall consumer prices
How did single purchases fare at different holding periods?
- 20-year windows that preserved purchasing power
- Median 20-year purchasing-power change
So... time alone isn't enough?
- Long holding periods could still lose buying power. Preservation rates differ across durations and historical samples; compare the common-start results below before attributing those differences to time alone. Even some 25-year purchases lost buying power; these historical frequencies are not future probabilities.
- A difficult purchase date can matter for decades. The 1980–2000 example lost substantial purchasing power despite a 20-year wait. A higher dollar price alone also doesn't establish that gold kept up with your cost of living.
A pooled history of monthly purchases
Suppose you bought $100 of gold every month and held each purchase for the same number of years. This spreads purchases across different prices. We combine every completed purchase and its later sale to measure the change in purchasing power.
- Net purchasing-power change
- After-cost buying power per dollar contributed
Each purchase gets its full holding period. The totals combine sales across different dates; they are not a balance available on one date.
What can monthly buying tell us?
- Your purchases span many different prices. Investing a fixed $100 each month buys more gold when its price is lower and less when it is higher. Your combined result comes from many purchase dates instead of depending on a single one.
- A pooled result does not measure a typical investor. Buying and selling costs and taxes can change whether the strategy preserves purchasing power. Combining decades of purchases can hide unsuccessful starting periods. These results do not isolate the benefit of the buying schedule or establish that monthly buying beat a lump sum available at the outset.
How do the pooled purchases compare with an S&P proxy?
The same pooled $100 monthly purchases, held for the same number of years, in a historical S&P Composite proxy. Compare gold with a stock-market proxy that includes reinvested dividends, assumed taxes, brokerage fees, and fund expenses.
This comparison needs JavaScript and the saved monthly stock data.
- Gold net purchasing-power change
- S&P net purchasing-power change
What if you started buying monthly at a different time?
A monthly investor builds one portfolio and eventually sells it. Here, each possible starting month gets its own simulation: buy $100 of gold each month for the selected duration, then sell the accumulated holdings together. Recent purchases therefore have shorter holding periods than early purchases.
Uses the buying premium, selling discount and long-term gold tax assumption in the pooled section above. Gold lots held 12 months or less use the short-term rate here. T-bill interest is taxed monthly at the entered rate. Rates are constant illustrative assumptions.
- Investor windows preserving buying power
- Median net purchasing-power change
Compared with cash and Treasury bills
All three receive the same monthly $100 contributions. The Treasury-bill proxy compounds published monthly returns; idle cash earns zero interest. Monthly-average gold prices and monthly bill returns have different observation conventions, so this is an approximate comparison rather than executable trading results.
Does the holding-period comparison change with the same start dates?
Longer periods normally have fewer recent starting dates. This comparison keeps only purchase months with complete 1-, 5-, 10-, 15-, 20- and 25-year endpoints, so every duration begins with exactly the same historical sample. Sale dates still differ.
These are overlapping single-purchase windows, not monthly-investor portfolios. Holding the start-date sample constant reduces one source of bias; it does not isolate a causal effect of holding longer.
Did monthly investors fare differently across inflation regimes?
Using the investor duration and costs selected above, group the completed windows by annualized overall CPI inflation from the first purchase to liquidation. Thresholds are descriptive choices: below 2%, 2% to below 4%, and 4% or more.
These groups use inflation observed over the entire window, which was unknown at the start. They describe historical associations, not a timing signal or proof that gold reliably hedges an inflation surprise. Long windows can contain several very different inflation episodes; overlapping windows also repeat those episodes.
So what does this tell us?
- Inflation erodes the buying power of idle dollars. As consumer prices rise, the same amount of cash buys less. That erosion compounds over time; cash needs to earn enough after taxes to keep pace with inflation.
- Gold's outcome depends on when and how you invest. Single-purchase distributions, pooled purchases, and rolling investor portfolios answer different questions. Use the failure rates and worst outcomes alongside the medians; a positive full-history total cannot establish that a typical starting date succeeded.
- Monthly buying spreads entry dates, but does not guarantee preservation. Costs, nominal-gains taxes, inflation and the eventual sale date all matter. These simulations do not compare monthly investing with investing an available lump sum, or measure portfolio diversification.
- Use an interest-bearing alternative when comparing with cash. Idle dollars and Treasury bills have different outcomes. The rolling comparison uses matching contributions and dates; the separate stock comparison is a pooled historical proxy. None establishes a guaranteed future ranking or a recommendation for your portfolio.
Sources and calculation
Monthly data from January 1960 through August 2026. Each category uses its own BLS price index and available history. Missing observations are excluded from selections and shown as chart gaps.
Both lines begin at your purchase amount. Gold above the benchmark means greater buying power than at purchase.
Gold is the monthly average USD price per troy ounce, from the DataHub gold dataset, using World Bank commodity price data from 1960 onward. Earlier annual figures repeated as monthly values are excluded.
Purchasing power uses BLS CPI-U indexes for the U.S. city average, not seasonally adjusted. Select the broad consumer basket or a specific category. Your personal prices and location may differ from these national averages. The snapshot uses BLS observations published by FRED.
- Overall consumer prices — the broad U.S. consumer basket, including goods and services; all items, CUUR0000SA0.
- Housing costs — CUUR0000SAH; shelter, utilities, and household furnishings and operations. Available from January 1967. This does not measure house purchase prices.
- Rent — rent of primary residence, CUUR0000SEHA.
- Groceries — food at home, CUUR0000SAF11.
- Gasoline — all types, CUUR0000SETB01; excludes other energy costs.
Most categories have no observation for October 2025 because of the lapse in appropriations. Gasoline also has gaps in 1964–1966. No missing values are estimated.
Gold acquired = purchase amount ÷ purchase gold price. Ending value = ounces × ending gold price. Category benchmark = purchase amount × (ending category index ÷ purchase category index).
Purchasing-power change = [(ending gold price ÷ purchase gold price) ÷ (ending category index ÷ purchase category index) − 1] × 100. See the BLS explanation of purchasing power.
The historical sections always use overall CPI-U, independently of the scenario controls. The counterexample starts with $100 in January 1980. Holding periods of 1, 5, 10, 15, 20, or 25 years start in each available month from January 1960 and end exactly the selected number of calendar years later; both endpoints must have gold and CPI observations. Missing endpoints are excluded, with no interpolation. Preserved means a purchasing-power change of zero or greater. The median is the middle total purchasing-power percentage change when all outcomes are sorted; for an even number of windows, it is the average of the two middle changes. It is not an annualized return. Windows overlap and are not independent trials or a forecast. Chart bands adjust to the range of outcomes for the selected period; each band includes its lower boundary and excludes its upper boundary.
The monthly strategy invests a fixed nominal $100 at each monthly average gold price, then sells that purchase exactly the selected number of calendar years later. Only completed pairs with both CPI observations are included; missing pairs and purchases still awaiting their sale date are excluded. Contributions and sale proceeds are each converted to the latest available overall-CPI month's dollars: amount × (reference CPI ÷ CPI at that transaction). Combined purchasing-power change = (sum of adjusted proceeds ÷ sum of adjusted contributions − 1) × 100. The multiple is the same ratio before subtracting one. These are pooled totals, not a median, an annualized return, or a portfolio balance. Proceeds are not reinvested; cash between transactions earns no modeled interest. Different holding periods include different purchase-date ranges.
The comparison assumes gold is bought and valued at monthly averages. The initial calculator, counterexample, and individual-window distribution exclude dealer premiums, selling spreads, storage, insurance, and taxes. The monthly strategy includes the assumptions shown beside its controls. It compares with price changes in the selected category; the benchmark is not an interest-bearing investment.
The monthly strategy uses a fixed $100 all-in budget with the entered buying premium and selling discount (defaults: 3% and 5%). Ounces = $100 ÷ [purchase gold price × (1 + premium ÷ 100)]. Sale amount = ounces × ending gold price × (1 − discount ÷ 100). The $100 cost basis includes the buying premium. Modeled tax on each purchase = max(0, sale amount − $100) × the entered tax rate. Net proceeds = sale amount − modeled tax. Only then are proceeds inflation-adjusted to the reference month and summed. The displayed monthly results and chart include these costs and taxes.
The tax field is a simplified sensitivity assumption applied to every historical sale, not a reconstruction of historical tax law or a personal tax calculation. Physical gold held more than one year is generally subject to the collectibles rules, with a maximum federal capital-gains rate of 28%; the rate can be lower. The 1-year option is short-term and generally subject to ordinary income-tax rates, so use an appropriate rate in the field. See IRS capital-gains guidance and Publication 550. The model taxes nominal gains, not inflation-adjusted gains. It does not model capital-loss offsets, carryovers, state taxes, the net investment income tax, tax-payment timing, or changing tax rates. Storage, insurance, shipping, and payment charges are also excluded.
Premiums and discounts are illustrative, not universal quotes. Actual terms vary: JM Bullion explains premiums and product size, and APMEX explains product-dependent buyback prices. Small physical pieces can carry higher premiums; a fractional $100 monthly purchase at these terms is a modeling assumption.
Dollar-cost averaging means investing equal amounts at regular intervals, as described by Investor.gov. The pooled calculations and rolling investor calculations answer different questions; neither isolates the effect of the purchase schedule from gold’s price history and the holding period.
The pooled monthly section combines one full-history set of completed purchases for each holding period. The rolling section separately tests every fully observed monthly-investor start. Neither compares with lump-sum investing, tests portfolio diversification, or predicts future results. For additional context, Erb and Harvey’s The Golden Dilemma finds gold an unreliable inflation hedge over practical investment horizons.
The stock comparison uses Robert Shiller’s S&P Composite monthly dataset, from 1960 onward. Prices are monthly averages; the dividend column is an annualized amount derived from quarterly totals with monthly interpolation. We approximate monthly dividends as that amount ÷ 12, paid at each subsequent monthly observation. This is a fund proxy, not the realized returns of a specific fund available throughout the full history. Months lacking published dividends are excluded; stock-source CPI estimates are never used. The comparison uses our same BLS overall CPI and exactly matched gold/stock purchase and sale dates.
The stock holding period follows the monthly-gold toggle above. The percentage brokerage fee is deducted from each $100 all-in purchase and each sale; no brokerage commission is charged on modeled dividend reinvestments. An annual expense rate is applied to the proxy fund NAV as (1 − expense rate)^(elapsed months ÷ 12). Each month, dividend tax is withheld and only the remaining dividend is reinvested at the proxy NAV. Reinvestments increase cost basis and are tracked as separate lots. At the ending sale, each lot’s nominal gain, after selling fees, is taxed at the entered long-term rate if held more than 12 months and the short-term rate otherwise. Exactly one year is short-term. No tax benefit is credited for losing lots. Net proceeds are then adjusted into the same reference-month dollars as the contributions and gold proceeds.
The stock defaults are illustrative current-style assumptions: 15% long-term gains tax, 24% short-term gains tax, and 15% dividend tax. Qualified dividends can use the preferential federal rates, but not every distribution is qualified; the model applies the entered dividend rate to all distributions without testing qualification. See IRS capital-gains rules and dividend rules. Rates are held constant throughout historical simulations, rather than reconstructing past tax law. State taxes, NIIT, loss netting, carryovers, tax-payment timing, fund capital-gains distributions, tracking error, and bid/ask spreads are excluded.
The 0% brokerage default reflects commission-free online U.S. stock/ETF trading, such as Fidelity’s published pricing; regulatory selling fees and other charges can still apply. The 0.03% expense default is based on the published Vanguard S&P 500 ETF expense ratio. These assumptions are applied to the historical index proxy; they are not a claim that the same fund or pricing existed since 1960.
Rolling investors contribute exactly $100 at the monthly average gold price in months 0 through N−1 and liquidate all lots in month N, where N is duration in years × 12. Each lot uses the premium and discount above; positive nominal gains are taxed separately at the short-term rate when held at most 12 months and the pooled gold tax rate otherwise. No loss offsets are credited. Contributions are converted individually into liquidation-month dollars using that month's CPI divided by contribution-month CPI. Net purchasing-power change is total net liquidation proceeds divided by the sum of these adjusted contributions, minus one. It is neither an annualized return nor a comparison against investing all contributions on day one. Every contribution month and liquidation month must have gold and CPI data; no missing observations are filled. A CPI gap can therefore exclude many overlapping investor windows. The reported preservation share, median and extremes weight each complete starting month equally.
The Treasury-bill proxy uses the monthly RF column from Kenneth French's Data Library. Its construction notes identify one-month Treasury-bill data from Ibbotson Associates through May 2024 and the ICE BofA US 1-Month Treasury Bill Index from June 2024. These are published monthly percentage returns, converted to decimal returns, rather than annual yields divided by 12. Each deposit starts earning the next month's return, including the final liquidation month. Positive interest is taxed at the entered rate monthly; negative returns receive no tax benefit. This is a simplified tax-timing convention; no fees are modeled. This is a historical proxy, not a specific account, executable bill ladder, or guaranteed after-tax return. Comparisons require all intervening monthly returns; gold is recalculated on the exact bill subset. Both series use the same CPI-adjusted contributions. The stock comparison above retains its separate pooled methodology.
Common-start comparisons retain only single-purchase start months with valid gold and overall CPI at all six exact calendar endpoints. Inflation-regime groups use realized annualized endpoint CPI change, (ending CPI / starting CPI)^(1 / years) − 1, over each rolling investor window; their gold metrics include the rolling costs and taxes. These are retrospective, overlapping groups without significance tests. Gold's pre-1971 fixed-price history and historical U.S. ownership restrictions also limit how an investor should interpret the full sample as an investable strategy.
The site uses a saved historical snapshot, refreshed separately from source data. It includes only completed months available in both series. Results use unrounded source values; displayed amounts are rounded.