What B.C.’s Property Tax Changes Can Teach Us About Land Value Tax
By Liam Wilkinson
TLDR
Our first major B.C. modelling result gives us an important empirical benchmark for the next stage of our Land Value Tax work.
Higher recurring holding costs appear to be capitalized into land prices over time, rather than showing up as a dramatic one-time speculative crash.
Properties subject to the AST and SVT did not appear to reduce productive investment, and our broader preliminary work suggests stronger improvement activity in the treatment group.
We are seeing early signs of redevelopment effects, including movement from single-family toward multifamily use, changes in building age consistent with teardown and rebuild activity, and increases in bedrooms.
Most importantly, these findings help us calibrate the larger B.C. LVT model, test whether its outputs track observed market behaviour, and better estimate housing elasticity in B.C. markets.
There is no clear visual evidence that AST implementation reduced the median improvement share of the ≥$4M group; if anything, the series appears modestly higher after 2019.
In plain terms, this work helps us move from theory toward measurement. It gives us a better basis for setting the parameters that drive our model’s long-run projections, including expected effects on land prices, redevelopment, housing supply, and the intensity of land use. It also gives us a way to test our assumptions against real-world outcomes and see how closely the model predicts what actually happened under a partial increase in carrying costs.
We have now completed the first major stage of our B.C. modelling work using a natural experiment created by recent provincial property tax changes, especially the Additional School Tax and, more broadly, the Speculation and Vacancy Tax framework.
These are not pure Land Value Taxes, and we are not treating them as such. But they do give us a useful real-world test of something central to our broader project: what happens when government increases the recurring cost of holding valuable residential property.
That matters because our objective is not simply to argue for Land Value Tax in principle. It is to build a comprehensive model for British Columbia that can estimate how different LVT designs might affect land prices, redevelopment, housing supply, and land use over time. To do that credibly, we need empirical evidence that helps us calibrate the model’s behavioural assumptions and evaluate whether its outputs line up with observed outcomes.
-
For readers new to this work, we are building a detailed Land Value Tax model for British Columbia in order to test how different policy designs would likely play out in practice.
That includes questions such as: how much would land prices adjust, how quickly would those effects appear, how much redevelopment might follow, how sensitive B.C. housing markets are to these changes, and whether higher carrying costs would translate into more units, more bedrooms, or more intensive land use.
A major part of that task is estimating housing elasticity in B.C. markets. In other words, we need a better sense of how strongly housing supply and redevelopment respond when incentives change. This natural experiment helps us do that. It gives us observed market behaviour that we can use both to calibrate the model and to test how closely its projections match empirical results.
Why B.C. gave us a useful natural experiment
One of the hardest parts of modelling an LVT is that the key effects run through behaviour.
If the cost of holding land rises, do owners sell, redevelop, add units, change ownership structures, or try to avoid the tax? How much of the burden shows up in lower land values? How much shows up in changes to land use? And how much varies depending on local supply conditions?
British Columbia gave us a useful opportunity to begin answering those questions.
In 2018, the province announced the Speculation and Vacancy Tax, an annual tax aimed at discouraging vacant homes and housing speculation in designated regions. It then introduced the Additional School Tax, which applied from 2019 onward to the residential portion of properties assessed above $3 million, with a higher marginal rate above $4 million. These policies are not equivalent to a pure LVT, but they do increase the annual carrying cost of holding certain residential properties, which makes them relevant to the mechanisms we are trying to model.
The AST analysis in our write-up uses historical property assessment data in Vancouver and compares properties just above the tax threshold to similar properties just below it. That makes it a useful natural experiment. The treated and control groups were tracking closely before the tax came into force, which gives us a stronger basis for attributing subsequent divergence to the policy change rather than unrelated market noise.
What we found
The first major finding is near-perfect capitalization.
That is a technical term, but the underlying question is straightforward: when the recurring cost of holding a property rises, how does that cost get reflected in the price of the property itself?
Our early answer appears to be, slowly and completely.
In our analysis, we did not find evidence of a sharp price discontinuity right at the $3 million or $4 million thresholds. Regression discontinuity estimates show no statistically significant immediate break in land values at those cutoffs. But over time, the pattern is clearer: taxed properties experienced a statistically significant reduction in annual land value growth of roughly 1.5 to 2.0 percentage points after 2019 relative to comparable properties below the threshold.
That suggests the effect is showing up through slower land appreciation rather than through a dramatic one-time repricing. For our purposes, that is important because it gives us a more grounded estimate of how carrying costs feed into land values over time, which is exactly the sort of relationship our broader model needs to capture.
The second major finding is that productive investment does not appear to fall.
We find no evidence that taxed properties reduced their relative investment in structures or renovations after the tax took effect. Improvement shares among taxed properties track closely with comparable untaxed properties and in some years slightly exceed them.
Our broader preliminary work points in the same direction. We found a significant increase in improvement values for properties subject to these taxes relative to those that were not. For the school tax, we observed roughly 18% higher improvement values over the control group, with an effect that appears persistent.
That does not settle every question, but it does suggest that higher carrying costs did not suppress productive use in the way critics often assume. If anything, the pattern is more consistent with a shift away from passive holding and toward improvement or intensification.
We are also seeing early evidence of redevelopment effects.
In the treatment groups, there appears to be an increase in zoning changes from single-family toward multifamily use, although we still want to run additional controls before speaking too firmly about the precise magnitude. That interpretation is supported by changes in building age, which are consistent with teardown and rebuild activity, and by increases in the number of bedrooms, which may reflect added units or ADUs.
These are particularly useful findings for the next stage of the modelling because they help us estimate the responsiveness of the housing stock itself. That includes not just prices, but housing elasticity in a broader sense: how strongly landowners respond to changed incentives by altering the built form, increasing capacity, or shifting toward more intensive use.
Why this matters
The main significance of this result is methodological.
It gives us empirical evidence we can use to calibrate the larger B.C. Land Value Tax model, rather than relying too heavily on theoretical assumptions. It also gives us an opportunity to test the model itself by asking whether its outputs track observed market responses under a real-world increase in carrying costs.
That matters for several reasons at once. It helps us refine the model’s assumptions about capitalization. It gives us a better basis for estimating redevelopment responses. And it helps us properly calibrate housing elasticity in B.C. markets, which is central to any serious attempt to forecast long-run supply effects.
So while this result is preliminary, it is useful in a very practical way: it improves both the model’s inputs and our ability to judge the model’s accuracy.
Where we go from here
This is the first major result of many.
The next step is to tighten and extend the analysis. We want to add more controls around the redevelopment and zoning findings before drawing stronger conclusions about magnitude. We want to separate genuine supply responses from tax-planning responses more clearly. And we want to examine how these effects vary across neighbourhoods, policy designs, and property types.
All of those results feed back into the full B.C. model.
Each finding helps us set the behavioural parameters that determine how the model projects changes in land values, redevelopment, housing supply, and land use over time. Just as importantly, they help us evaluate whether the model is producing outputs that are consistent with observed empirical outcomes.
That is a more useful standard than simply asking whether the theory sounds plausible. The question is whether the model behaves in a way that tracks the world closely enough to be informative for policy design.
At this stage, the evidence is encouraging. It suggests that higher carrying costs are associated with slower land price growth, continued or stronger improvement activity, and early signs of more intensive land use. It also suggests that any serious policy design will need to account for behavioural adaptation and avoidance.
That is exactly the sort of evidence we hoped to generate at this stage of the project with much more to follow.
Stay informed about this work by signing up for our Common Wealth Canada email list.
