← Back to all insights

How long does an average worker have to work to afford one ounce of gold? In August 2026 the answer was about 117 hours, close to three working weeks.

That is more than three times the average of the last 100 years (33 hours), and in the US data since 1920 only five months in early 2026 were higher. It is still far below the 500-year average (about 760 hours): in 16th and 17th century England it took roughly 900 to 2,400 hours.

Switch to the inflation rate to see the same history through consumer prices. Scroll or pinch to zoom, drag to pan, and open Sources for the data behind every line.

Treat the older numbers with care. Everything before 1257 is a rough estimate from scattered records, and US figures before 1964 are manufacturing pay scaled to today's definition.

What the trend suggests

The long fall from the 1200s to 1970 fits what economic historians describe. Clark (2005) finds that English real wages, what pay buys in goods, were trendless before about 1800 and rose afterwards. For much of this period the price of gold was fixed by law (£4.25 an ounce in Britain from 1717, $35 in the US from 1934), so the hours needed fall whenever money pay rises. That makes the fall a rough proxy for rising pay and productivity, though not a clean measure, because money pay also rises with inflation.

Since 1970 the line has turned up: 8.5 hours in August 1970, 86 in January 1980 and 117 today. In August 1971 the US ended the convertibility of the dollar into gold and the gold price began to float (Federal Reserve History). From 1970 to August 2026 gold rose about 123 times in dollars, average hourly pay about 9 times and consumer prices about 9 times. Gold therefore gained more than 13 times on both pay and prices.

Is that a loss of wealth? In terms of gold, yes: an hour of work buys about 13 times less gold than in 1970. In terms of goods the picture is mixed. On this chart's pay series, hourly pay after consumer prices is up only about 7% in 56 years, and the series counts wages only, not benefits. Stansbury and Summers (2017) show that median pay has diverged from productivity since 1973, although productivity still feeds through in part: one extra point of productivity growth goes with 0.7 to 1 point of median and average pay growth.

Gold is also a poor yardstick on its own. Its price in hours of work swings: from 86 in 1980 to 15 in 2001, and up again since. Jastram's "golden constant" holds that gold keeps its purchasing power over centuries. Erb and Harvey (2013) find that gold may hedge inflation over centuries but is an unreliable hedge over practical horizons, with below-average real returns after periods when its real price was high.

So the common reading, that a rising line means lost value, holds in part. A falling line has gone together with rising pay and productivity. A rising line that began when gold was set free reflects three things at once: a dollar that lost ground against gold, typical pay that lagged productivity, and gold's own boom and bust. It shows that wages buy far less hard money than they did, which is a real loss in that sense, but it is not a full measure of wealth.

Research and data