Ah Estonia, any time you look into national administration, they get a shout out for being great (you can even get divorced online there now). I visited Tallinn in 2022 and it was pretty sweet, hung out around Kalamaja, plus I bought a fluffy hat I still wear every winter. So when I came across this report of Estonia having the world's most competitive tax system, with LVT getting a mention, it felt like a good excuse to look closer.

Estonia is the only EU country that taxes land only, with buildings completely exempt: a pure land value tax in the Georgist sense. They've had it since 1993, shortly after independence, making it one of the longest-running real-world experiments in the idea. In true Estonian fashion, there's wonderful, clear documentation online about their land value tax and how valuations are done. Here are the key points.

How Estonia's LVT works

Rates are set locally, within national bands. Municipalities choose their rate within ranges set by national law: 0.1–1% of taxable land value per year for residential land, and up to 2% for other uses. In practice most municipalities land around 0.5–1%. This is considerably lower than Georgists typically advocate. The classic argument is that rates should be high enough that land speculation becomes unprofitable, which would require something significantly higher.

Per-plot automated valuation is new as of 2022. Previously, the mass valuation produced value zones, which local governments then used to calculate individual plot values. The 2022 law change shifted this: the national assessment now directly determines a taxable value for each cadastral unit using automated statistical models, drawing on actual transaction data, land use type, area, and proximity to water, with no on-site inspections. The one partial exception is the city centres of Tallinn and Tartu, where human appraisers set per-neighbourhood building-rights values, with automated calculation applied from there. The full methodology is publicly documented.

Valuations have happened in 1993, 1996, 2001, and 2022, with the next one due this year in 2026. The gap between 2001 and 2022 is the system's biggest practical weakness (more on that below).

The 2022 law change redesigned how valuations work, and the timing was urgent. Because land values had roughly increased sevenfold in the two decades since the 2001 assessment, a sudden update to accurate values would have caused a massive tax shock. The new law established the per-plot automated methodology, the first valuation under it was carried out later in 2022, and the results took effect in 2024, but are being phased in gradually. Increases were capped at 50% or €20 in 2025, and local governments can set caps of 10–100% per year from 2026 onward until assessed values are fully reached.

What can we learn?

The English-language research I found on Estonia's LVT points in a consistent direction: the system works directionally, but is too underpowered to deliver the full Georgist promise.

On speculation: A 2012 study found that Estonia did experience a significant speculative housing bubble in Tallinn between 2003 and 2008, and that a more rigorously implemented LVT would have acted as a buffer. The culprit wasn't the system design but the 2001 valuations being frozen in place: with no reassessment until 2022, the tax base drifted further from reality every year, reducing the anti-speculative effect.

On land utilisation: A study comparing Tallinn and Riga (Latvia kept a building-based tax) found more inner-city development and multi-family construction in Tallinn, a detectable positive signal. But sprawl was equally strong in both cities, and prime central Tallinn plots were still being used as paid parking lots rather than developed. The rate, at around 1%, is simply too low: parking revenue can still comfortably cover the tax cost, so the holding incentive isn't broken.

On economic efficiency: An analysis by the Organisation for Economic Co-operation and Development (OECD) found that raising land taxation to OECD norms while cutting labour taxes could lift Estonia's potential GDP per capita growth by up to 0.7%. This is a structural efficiency argument: land taxes don't distort productive behaviour the way income taxes do, because taxing a fixed supply of land doesn't reduce the amount of land available. The OECD's conclusion is essentially that Estonia has the right structure but not nearly enough intensity.

So what does this mean for Berlin?

The Estonian case is useful less as a success story and more as a proof of concept with a clear diagnosis of its own limitations. It shows that:

1. Automated per-plot mass appraisal at scale is now demonstrated. Estonia has just shown it can be done nationally using transaction databases and statistical models, without committees or on-site visits. This is directly relevant to Berlin, where Bodenrichtwerte (official land value maps) already exist as a regional data foundation. There's even research showing these can be carried out via statistical models, in a higher granularity.

2. Regular reassessment is not optional. The 2001–2022 gap turned a functional system into an ineffective one. The 2022 Land Valuation Act now legally mandates four-yearly valuation cycles, which is a vast improvement. Annual would be better still, the closer the tax base tracks actual land values, the less room for speculative drift, but four years is a reasonable starting point for a national system.

3. Rate matters enormously. A 1% LVT is simply not enough to change holding behaviour in valuable locations. Germany's new land value taxes are even lower in effective terms. If the goal is to make speculation unprofitable and idle land costly, the rate needs to be meaningfully higher, maybe in the 5-10% range to capture the full land rent values.

Estonia's experience doesn't prove LVT works perfectly. But it does suggest the mechanism works in the direction theory predicts, and that the failure modes are about implementation (stale valuations, low rates) rather than the underlying idea. Personally, I kind of wish they'd just jack it up to 11, do annual appraisals, much higher rates, lower other taxes and see what happens. For Berlin, the Estonian case shows that the data infrastructure needed for a well-functioning LVT isn't theoretical: automated per-plot valuation at city scale is demonstrably achievable.