A car in Greece loses about 8% of its value every year - and the rate barely changes with age

There is a sentence that turns up in every conversation about used cars: a car loses most of its value in the first few years and then levels off. We have written it ourselves. Measured across the whole Greek market, it turns out to be half true - and the missing half changes the arithmetic for anyone deciding when to sell.

The finding

Value loss in Greece is a near-constant percentage per year, not a steep drop that flattens out. A single rate - 7.9% a year - predicts every point on the curve, from one year old to twelve, to within two percentage points.

Why it looks like it flattens

The sense that the curve flattens is real, but it is about euros rather than the rate. A two-year-old car loses 7.2% of a large number. An eight-year-old loses 6.4% of a much smaller one. The percentage is almost the same; the amount is half. What an owner notices - "it hardly loses anything now" - is the base shrinking, not the rate easing.

Why the Greek curve is shallower

In northern Europe an eight-year-old car is usually a matter of a few thousand euros. Here it still holds more than half its value. Two reasons account for most of the gap, and both are local: registration tax keeps new cars expensive, and therefore everything priced beneath them, and the fleet is old, so there is genuine demand deep into the curve.

The second shows up in our own data. Of the 58,419 live listings we hold, 21% are cars over 15 years old, and the median listing is nine years old. Almost half the listings - 45% - are more than a decade old. In a market like that, value is not allowed to fall to nothing: there is a buyer.

How it was measured

The obvious method - median price by model year - gives the wrong answer, because the mix changes: older years are full of superminis and recent ones full of SUVs nobody was buying in 2015. A curve drawn through them measures what Greeks bought in which year, not what a car is worth now.

So each model is first compared against its own newest year, and each point on the curve is the median of those shares. A model with 4,000 listings cannot pull the curve any harder than one with 40. A model year counts only with at least 15 listings, and a point only with at least 8 models.

What this data does not show

These are asking prices, not sale prices. We hold no transaction data and do not pretend to.

The comparison is cross-sectional: it shows what different model years cost today. It does not follow individual cars over time.

And the first step is not the forecourt drop. It compares one model year against the next within the used market - not list price against resale. Only 11% of listings are cars under four years old, so the first-day loss is not measured here and should not be read into that 5.4%.

Using the data

The figures are free to reuse with attribution, under CC BY 4.0. No permission is needed and there is no embargo: republish the tables, build your own charts, or download the numbers as JSON. We only ask for a link.