Land Value Tax, Vertical Equity, and Fixing Property Tax in Europe with Rob

About this Episode

Rob is a statistician based in Belgium who came to property tax through an unlikely door: the land value tax. His entry point was a deceptively simple thought experiment — what happens when a child arrives in a world where all the land is already claimed? That question led him deep into the economics of property taxation, vertical equity modeling, and ultimately to building mass appraisal tools. In this conversation, Rob makes a case that property tax isn't just administratively important — it's foundational to economic efficiency and, potentially, to peace itself.

What emerges is a wide-ranging discussion that moves from Georgist philosophy to the absurdity of Belgium's 1975 valuations, from vertical equity corrections to the future of model validation. The throughline is clear: property tax is the most efficient tax we have, and we're badly underinvesting in getting it right.

The Philosophical Case Is the Practical Case

Rob draws a direct line from natural resource allocation to conflict. Ukraine is about land. Israel is about land. The Strait of Hormuz is about natural monopolies. His argument is that the Georgist framework — tax what nature provides, leave alone what labor and capital produce — isn't just morally tidy. It's the equilibrium that prevents violence.

This matters for assessors because it reframes property tax as something far more consequential than a local revenue mechanism. Rob points to the history of Asian economic development, where land reform preceded explosive growth. The land value tax, he argues, is simply a more sophisticated technology for achieving the same redistribution — without the catastrophic efficiency losses seen in Zimbabwe or South Africa when land was physically seized and redistributed to people without the infrastructure to manage it.

The takeaway for practitioners: the philosophical arguments for property tax aren't abstract. They're the reason the tax exists and the reason it should be defended.

Belgium: A Cautionary Tale in Stale Valuations

Belgium's property tax is, as a percentage of GDP, one of the largest in Europe. That sounds impressive until you learn the last real valuation was done in 1975. The regression formula from that era penalizes homeowners for having a second toilet — because fifty years ago, that signaled a villa. Today it signals a normal house.

Location values have shifted dramatically. Ghent has surged in desirability while Antwerp's relative position has changed, but the tax rolls reflect a world that no longer exists. Values are indexed to a consumer price index, which the courts have twice upheld as acceptable — but indexing doesn't capture shifting spatial relationships or evolving housing stock.

Perhaps most damning is the opacity. Rob describes a citizen who had to send ten signed letters and travel to another city for a single chance to view an Excel spreadsheet explaining his assessment. The system survives precisely because no one can see how it works. That's stability through obscurity, not through legitimacy. And it comes at a cost: because land values rise faster than building values over time, the effective tax increasingly falls on structures rather than location — punishing development and investment.

Vertical Equity: It's a Mean Reversion Problem

The vertical equity problem — overtaxing low-value properties, undertaxing high-value ones — is endemic to mass appraisal. Rob's explanation is disarmingly simple: every model, whether linear regression or gradient boosted trees, gravitates toward the mean. High-value properties get pulled down. Low-value properties get pulled up.

But Rob adds nuance beyond the statistical explanation. Larger parcels sell for less per square foot partly because they require more internal infrastructure, have proportionally lower transaction costs, and face less buyer competition. These are real market dynamics that models struggle to capture — and that policy choices like transfer taxes make worse by muddying the sales data.

Rob also raises a point that deserves more attention: no jurisdiction he's aware of uses progressive mill rates for property tax. Income taxes are progressive. Property taxes are flat. If vertical equity is truly the concern policymakers claim it is, the simplest fix might not be better models — it might be differentiated rates. He's not necessarily advocating for it, but the absence of the conversation is telling.

The modeling fix Rob developed involves identifying the consistent, systematic bend in the predicted-versus-actual curve and applying a second-order correction. Because the bias is consistent, it's correctable — but only if you're testing properly.

Cross-Validation: The Field's Blind Spot

Rob is emphatic that the biggest methodological gap in mass appraisal isn't exotic algorithms — it's basic model validation. Too many jurisdictions train a model on their sales data and then test it on the same data. That's not validation. That's confirmation.

Fivefold cross-validation — where every parcel gets a prediction generated without the model ever having seen that parcel — is computationally trivial with modern hardware. The excuses for not doing it have evaporated. And the stakes are high: property tax ratio studies rely on metrics like COD and PRD that are extremely sensitive to outliers. Testing on a lucky holdout set can mask serious problems.

This isn't frontier research. It's standard practice in every other field that uses predictive modeling. The fact that it remains uncommon in assessment should be uncomfortable for the profession.

Key Takeaway

Property tax is the most efficient tax available to any government, but its survival depends entirely on valuation quality and public trust. Belgium's fifty-year-old regression formula isn't just an embarrassment — it's a preview of what happens when jurisdictions stop investing in accuracy. The field doesn't need more complex models nearly as much as it needs the discipline to validate the ones it already has.