The year the government lost track of coffee prices
For most of 2018 and 2019, the official US price of a pound of coffee simply vanished. We dug into why, filled the gap with other data, and found a story about who benefits when bean prices crash.
While building our coffee price tracker, we hit a hole in the data. The government’s monthly average price for a pound of ground coffee runs back to 1980, but in 2018 it starts to stutter, and then it disappears. After May 2018 there’s a single reading in September, then nothing until October 2019.
For 16 months, apart from that one reading, the official US price of coffee just isn’t there. So what happened?
What’s actually missing
The figure in question is the Bureau of Labor Statistics’ average price of 100% ground roast coffee, in all sizes, across US cities. It’s the number we chart on the tracker, and it has three holes:
| Missing months | Length | Why |
|---|---|---|
| September 2007 to November 2009, except November and December 2007 | 25 months | Most likely too few eligible prices (see below) |
| June 2018 to September 2019, except September 2018 | 15 months | Most likely too few eligible prices |
| October 2025 | 1 month | Government shutdown: no prices were collected at all |
The last one has a simple explanation. During the federal shutdown from October 1 to November 12, 2025, BLS says it couldn’t collect October’s price data, and decided collecting it after the fact would be costly and potentially inaccurate. Every price series has a hole there, not just coffee.
The other two are more interesting.
Why the number vanished
When the BLS publishes “average prices”, they’re a by-product of the much bigger job of measuring inflation for the consumer price index (CPI). Data collectors record the prices of specific items in a sample of stores, and that sample changes over time: BLS says about a quarter of its sample rotates every year.
The catch is in BLS’s own words: the sample is “optimized for the calculation of price indexes, not for the calculation of average prices.” To publish an average price, BLS requires a minimum number of eligible price quotes. For coffee, that means a regular, caffeinated ground coffee in a can or plastic container. If too few of the items in the sample fit that description in a given month, the average isn’t published. BLS warns that as the sample rotates, a series can fall below that minimum “for several years or even permanently.”
That’s the most likely explanation for 2018 and 2019. As stores and products rotated through the sample, too few quotes matched the coffee definition, until enough did again in late 2019. We should be clear that BLS doesn’t single out coffee or these years in anything we could find. This is the mechanism BLS documents, not a specific explanation it has given.
Notably, the coffee price index never stopped. The CPI for coffee, which tracks how prices change rather than what they are, was published every month through both gaps, because it doesn’t need the same tight product match. That turns out to be the key to filling the hole.
Filling the gap
If we know the price before the gap and after it, and we know how much prices moved each month in between, we can make a good estimate of the missing months.
Here’s the method in plain terms:
- Take the last real price before the gap and move it forward month by month, in step with the coffee price index.
- Do the same backwards from the first real price after the gap.
- Blend the two, leaning on the nearer real price, so the estimate lines up with real data at both ends.
Before trusting it, we tested it. We took stretches of the record where BLS did publish prices, pretended 16 months were missing, and compared our estimates with the real figures. Across more than 2,200 test months between 2010 and 2025, the estimate was off by 2.8% on average and by less than 6% nine times out of ten. That’s close, but it’s still an estimate, so we’ve labeled it as one.
What US coffee probably cost in 2018–2019
| Month | Estimated price per lb |
|---|---|
| June 2018 | $4.29 |
| July 2018 | $4.30 |
| August 2018 | $4.31 |
| September 2018 (published) | $4.31 |
| October 2018 | $4.30 |
| November 2018 | $4.23 |
| December 2018 | $4.25 |
| January 2019 | $4.26 |
| February 2019 | $4.25 |
| March 2019 | $4.25 |
| April 2019 | $4.24 |
| May 2019 | $4.24 |
| June 2019 | $4.29 |
| July 2019 | $4.23 |
| August 2019 | $4.26 |
| September 2019 | $4.20 |
| October 2019 (published) | $4.17 |
The headline: almost nothing happened on the shelf. A pound of ground coffee cost about $4.30 when the data went dark and about $4.17 when it came back, a drift of roughly 3%. The coffee price index tells the same story, slipping only about 2% between January 2018 and October 2019.
Meanwhile, at the farm end
That calm is striking, because the global market for coffee beans was in turmoil.
The International Monetary Fund’s benchmark price for mild arabica, a widely used yardstick for quality arabica beans, fell from about $1.39 a pound in January 2018 to about $1.21 by May 2019, a drop of around 13%. Other benchmarks fell further: the International Coffee Organization’s composite price hit its lowest monthly average since 2006 in April 2019, and New York arabica futures dropped below $1 a pound, a 13-year low.
The cause was simple oversupply, with big harvests, especially in Brazil, for two years running. For many growers it was a crisis: by early 2019, farmers were warning that they were earning less than a cent for the beans in a cup of coffee, and many said prices no longer covered what it cost them to grow it.
Then, at the end of 2019, bean prices bounced back fast, up about 29% from the May low by December.
| January 2018 | Low point (May 2019) | December 2019 | |
|---|---|---|---|
| Global arabica beans, per lb | $1.39 | $1.21 | $1.56 |
| US ground coffee, per lb (estimated in the gap) | $4.29 | $4.24 | $4.05 |
Why your bag didn’t get cheaper
So the price of raw beans fell by about 13% and US shoppers saw almost none of it. Why?
- Beans are only part of the bag. Roasting, packaging, shipping, retail margins and marketing all sit on top of the bean price, and those costs didn’t fall.
- Contracts lag the market. Roasters typically buy months ahead, so a price drop takes a long time to reach the shelf, and short dips may never reach it at all.
- Prices are sticky downward. Retailers and brands tend to raise prices quickly when costs jump and lower them slowly, if at all, when costs fall.
The same lag can work in the other direction too: when bean prices spike, contracts signed earlier can delay the hit to shelf prices, and sticky pricing can keep them high after beans get cheaper. You can follow how that’s playing out today on the price tracker.
What we learned
The gap in the data turned out to be a gap in the story, too. In 2019, the people growing coffee went through one of the worst price slumps in a decade, while the price in American stores barely moved, and the government’s own yardstick for that price happened to be switched off.
About our estimates: they’re built from official BLS data, but they’re ours, not official figures. Our tracker still shows the real BLS series, gaps included. If you work with this data and know more about why the series was suppressed, we’d genuinely like to hear from you.