Underneath that national picture the crop also moved. The historic Eje Cafetero has lost more than a third of its share of national area since 2002, while Huila, Cauca and Nariño gained. The two stories turn out to be connected — and not in the direction you would guess.
Colombia's renovation programme was a response to a crisis. The 2008–2012 coffee leaf rust epidemic coincided with the country's worst production years in half a century — output fell to 7.7–7.8 million bags in 2011 and 2012 — and the FNC drove a mass replanting to rust-resistant varieties. The traditional park was almost entirely eliminated, falling from 147k hectares in 2007 to 4k in 2024.
The share view is where the turn is visible. Young technified area peaked at 85.9% in 2019. Every year since has been lower. Aged technified area, which had been driven down to 109k hectares, has risen back to 155k — a 43% increase in five years, with no interruption.
This is what a renovation wave looks like on the far side. Trees planted in one concentrated push do not age gradually across the park; they cross out of the "young" band together.
Between 2002 and 2025 the geography of Colombian coffee shifted decisively. In share terms the Eje Cafetero — Caldas, Quindío and Risaralda, the classic coffee axis — fell from 22.3% of national area to 13.8%. Huila, Cauca, Nariño and Tolima rose from 31.5% to 46.1%. Huila alone went from 78k hectares to 150k, and is now the single largest coffee department.
There are really two eras here, and conflating them is the easy mistake. From 2002 to 2013 the park grew, from 865k to 974k hectares, and the growth was overwhelmingly southern. From 2013 to 2025 it contracted by 141k hectares, and the contraction fell hardest on the old axis: Quindío lost 39% of its remaining area, Cundinamarca 33%, Valle 26%, Caldas and Risaralda 25% each. The migration had largely finished before the contraction began.
Here is the part that inverts the intuitive story. If you assume the shrinking regions are the neglected ones, you would expect the Eje Cafetero to hold the oldest trees. It holds the youngest.
Between 2019 and 2024 the aged share rose +5.6 percentage points in the South against +2.1 pp in the Eje Cafetero. The Eje is losing hectares, but what remains is comparatively well renovated. The South expanded fast between 2002 and 2013, and that expansion cohort is now aging as a block.
That is a concentration risk rather than a level risk. The national aged share of 18.6% is not alarming by itself. What matters is that the aging is bunched in one region and one planting vintage, which means the renovation requirement it implies will also arrive bunched — in the region that now carries 46.1% of national area.
The obvious next step is to ask whether an aging park predicts falling yield. We tested it, and the honest answer is that this data cannot establish it.
| Lag (years) | r — levels | n | r — first differences | n |
|---|---|---|---|---|
| +0 | +0.52 | 18 | -0.03 | 17 |
| +1 | +0.75 | 18 | +0.09 | 17 |
| +2 | +0.90 | 17 | +0.35 | 16 |
| +3 | +0.86 | 16 | +0.10 | 15 |
| +4 | +0.82 | 15 | -0.05 | 14 |
In levels, young share leads yield with a peak at lag +2 years (r = +0.90), which is physiologically the right shape — a coffee tree planted today bears in its second to third year. It is also almost certainly an artefact. Both series are strongly trended across only 18 observations, and two trended series correlate whether or not they are related.
First-differencing removes the shared trend. The lag-2 correlation falls to r = +0.35, and a 20,000-draw permutation test puts it at p = 0.18. Nothing is significant at any lag. The year-to-year changes do not confirm the relationship the levels suggest.
So: the aging is a measured fact, and the mechanism linking it to yield is plausible and well understood agronomically. But this dataset does not demonstrate the link, and we would need either a longer series or department-level panel data to test it properly.
For scale, Colombia's rolling twelve-month production peaked at 18.0 million bags in 1992-09. The latest reading is 12.7 million — about 29% below that mark, on a park that is 14% smaller than its 2013 extent and no longer getting younger.
Yield has done most of the work in the meantime: 17.3 bags per hectare in 2019 against 13.4 in 2002. That is the renovation dividend. The question the next few years will answer is whether it holds as the cohort that produced it ages — and whether Colombia renovates the southern wave before it has to.
Coffee area, production and price series — alongside the weather, positioning and balance-sheet data we track across origins — are queryable directly and through our MCP layer for AI-assisted analysis. If you would rather test the renovation-and-yield question yourself than take our null for it, that is exactly what the data is for.
Explore the Data