This all makes complete sense. Thanks for laying it out so clearly.
For anyone unconvinced, this is further reason to focus analysis of housing affordability exclusively on rents.
Since asset prices are buffetted around by interest rates and expectations and are skewed across jurisidictions by tax rates, the only reliable way to conduct like-for-like comparisons or comparisons through time is to use rental prices for dwellings of a consistent quality.
Thanks Cam. I agree the tax regime matters when comparing house prices across jurisdictions. For instance, Glaeser and Gyourko (2018) 'The Economic Implications of Housing Supply' incorrectly identified Houston as being a city where prices were below the 'minimum profitable production cost' (ie, a city in decline), when it's the classic fast growing city with responsive housing supply. That's because Texas makes above average use of land taxes to fund public works, so its house prices would've seemed too low compared to their default benchmarks.
But I wouldn't apply Texas's property tax rate to a Brisbane priced home. Their house prices are vastly less, so they pay less tax than your example implies. Also, maybe half of those taxes pay for schools and their income tax rates are correspondingly lower. Eg an example I have from 2018 is a new build in a Houston municipal utility district valued at $185k USD with annual taxes of $5.4k, of which $2.4k are for the schooling district. Given you won't pay for schools through your property tax, the $3k non-school property taxes are on par with what you pay.
If houses were trebled in price, all else equal, it's more reasonable to presume their tax rates would reduce to one third to maintain govt revenues. Also, your net income would be higher with Texas taxes because your income is not used to pay for schools serves your broader point thought: it's hard to compare house values across different states and countries.
Also in some states they cap use a much lower "assessed value" basis to apply the tax rate to, which is based on a price circa 1990 that increases only at something near the rate of general inflation (eg 2% in California, or 3% in Oregon). The market value and 'assessed value' can be very different. So US states generally wouldn't ever apply a 1.9% tax to an overpriced house such as in Brisbane, so it's not the right comparison. Perhaps if you reduced the value of your home to 1990 prices and then applied the 1.9% tax you might get it closer to the annual property taxes paid.
1. This is static in time analysis, however property taxes going up will reduce capital gains in the future vs a single stamp duty. (Just saw you mentioned this in the 2nd note).
2. This is still a good thing as it encourages mobility, allows people to downsize easier, or move to where they need to have jobs. (Having a land tax versus a stamp duty that is).
3. How does this work when analysing vacant land? Investors hold vacant land in the hope of appreciation in the future. It earns no yield or cashflow (in fact has a negative cashflow after holding costs). One would assume that a land tax would still be superior to encourage further supply and development of vacant land and thus result in better utilization.
4. The second house still appears "easier" to buy if using debt. Consider the math on a leveraged investment.
Buying the first house with 1302.4k of debt at say a 6.1% interest rate. = 79.4464k interest per year.
Plus after tax rent of 33.4k. Equals a shortfall of ~46.05K per year.
Buying the second house with 806.45k of debt at a 6.1% interest rate = 49.1935k interest per year.
Plus after tax rent of 20.677k. Equals a shortfall of ~28.52k per year.
A significant difference in cost of ownership. One cannot ignore the impacts of leverage on the housing market as it is a highly collateralizeable asset along with future potential growth in land scarcity and rents (therefore potential capital gains making up the shortfall). Making it easier for leveraged buyers to afford the house.
It appears to me that there is more than meets the eye in the interplay of all of these effects.
Overall good points, but I would argue you have overlooked one very important point on housing affordability this relates to.
In your example of your house dropping from $1.25m to $806k, it would reduce a 20% deposit for first homebuyers (who have far less equity/wealth and are more dependent on debt) from $250k to $161k, a 35% reduction ($89k is no small sum). If we took that down to a 10% deposit (or some of the 5% govt loan or shared-equity schemes available), that takes the deposit down to $80k, or even $40k - compared to $125k or $62.5k.
That's significant and would shave off a lot of time for first homebuyers, given the major homeownership barrier (at least prior to the recent interest rate rises) was the required time to save a deposit. Barriers to mortgage servicing only recently became the bigger issue with higher interest rates, but over the longer term as rates continue their decline I suspect we will see the deposit gap re-emerge as the dominate issue.
The growing problem we are facing is housing has become so much of an investment vehicle, we now require buyers to pre-save vast amounts of money before they can afford to purchase the asset. Yet they don't need that, they predominantly need the housing services and tenure benefits that homeownership provides.
Point is, how would we know if rents would rise, we don't know. We also don't know what changing property taxes will do with any certainty. We do know that the price of housing has gone stratospheric and that can't be based on any input costs. I know you have published some comparisons of different jurisdictions but it's difficult to draw any firm conclusions about these relationships from purely econometric data. We need a model that takes everything into account and fits the data in multiple jurisdictions. Your calculation is extreme hard to interpret in isolation.
Based on the assumption that house prices are determined by the marginal borrower, removing stamp duty would just push the price up because borrowed funds would be diverted from the tax to bidding up house prices.
How sure are we that house prices are determined by "yield"? What if prices are pushed up by financialisation, speculation and bank lending, and rents rise as a result?
Property taxes reduce housing asset prices but don't make housing cheaper
This all makes complete sense. Thanks for laying it out so clearly.
For anyone unconvinced, this is further reason to focus analysis of housing affordability exclusively on rents.
Since asset prices are buffetted around by interest rates and expectations and are skewed across jurisidictions by tax rates, the only reliable way to conduct like-for-like comparisons or comparisons through time is to use rental prices for dwellings of a consistent quality.
Thanks Cam. I agree the tax regime matters when comparing house prices across jurisdictions. For instance, Glaeser and Gyourko (2018) 'The Economic Implications of Housing Supply' incorrectly identified Houston as being a city where prices were below the 'minimum profitable production cost' (ie, a city in decline), when it's the classic fast growing city with responsive housing supply. That's because Texas makes above average use of land taxes to fund public works, so its house prices would've seemed too low compared to their default benchmarks.
But I wouldn't apply Texas's property tax rate to a Brisbane priced home. Their house prices are vastly less, so they pay less tax than your example implies. Also, maybe half of those taxes pay for schools and their income tax rates are correspondingly lower. Eg an example I have from 2018 is a new build in a Houston municipal utility district valued at $185k USD with annual taxes of $5.4k, of which $2.4k are for the schooling district. Given you won't pay for schools through your property tax, the $3k non-school property taxes are on par with what you pay.
If houses were trebled in price, all else equal, it's more reasonable to presume their tax rates would reduce to one third to maintain govt revenues. Also, your net income would be higher with Texas taxes because your income is not used to pay for schools serves your broader point thought: it's hard to compare house values across different states and countries.
Also in some states they cap use a much lower "assessed value" basis to apply the tax rate to, which is based on a price circa 1990 that increases only at something near the rate of general inflation (eg 2% in California, or 3% in Oregon). The market value and 'assessed value' can be very different. So US states generally wouldn't ever apply a 1.9% tax to an overpriced house such as in Brisbane, so it's not the right comparison. Perhaps if you reduced the value of your home to 1990 prices and then applied the 1.9% tax you might get it closer to the annual property taxes paid.
I would instead stress that the higher land tax does *not* make the house *more expensive* to rent (contrary to what all landlords claim).
Headline should say "... but don't make housing cheaper **to rent**"
A few thoughts.
1. This is static in time analysis, however property taxes going up will reduce capital gains in the future vs a single stamp duty. (Just saw you mentioned this in the 2nd note).
2. This is still a good thing as it encourages mobility, allows people to downsize easier, or move to where they need to have jobs. (Having a land tax versus a stamp duty that is).
3. How does this work when analysing vacant land? Investors hold vacant land in the hope of appreciation in the future. It earns no yield or cashflow (in fact has a negative cashflow after holding costs). One would assume that a land tax would still be superior to encourage further supply and development of vacant land and thus result in better utilization.
4. The second house still appears "easier" to buy if using debt. Consider the math on a leveraged investment.
Buying the first house with 1302.4k of debt at say a 6.1% interest rate. = 79.4464k interest per year.
Plus after tax rent of 33.4k. Equals a shortfall of ~46.05K per year.
Buying the second house with 806.45k of debt at a 6.1% interest rate = 49.1935k interest per year.
Plus after tax rent of 20.677k. Equals a shortfall of ~28.52k per year.
A significant difference in cost of ownership. One cannot ignore the impacts of leverage on the housing market as it is a highly collateralizeable asset along with future potential growth in land scarcity and rents (therefore potential capital gains making up the shortfall). Making it easier for leveraged buyers to afford the house.
It appears to me that there is more than meets the eye in the interplay of all of these effects.
Overall good points, but I would argue you have overlooked one very important point on housing affordability this relates to.
In your example of your house dropping from $1.25m to $806k, it would reduce a 20% deposit for first homebuyers (who have far less equity/wealth and are more dependent on debt) from $250k to $161k, a 35% reduction ($89k is no small sum). If we took that down to a 10% deposit (or some of the 5% govt loan or shared-equity schemes available), that takes the deposit down to $80k, or even $40k - compared to $125k or $62.5k.
That's significant and would shave off a lot of time for first homebuyers, given the major homeownership barrier (at least prior to the recent interest rate rises) was the required time to save a deposit. Barriers to mortgage servicing only recently became the bigger issue with higher interest rates, but over the longer term as rates continue their decline I suspect we will see the deposit gap re-emerge as the dominate issue.
The growing problem we are facing is housing has become so much of an investment vehicle, we now require buyers to pre-save vast amounts of money before they can afford to purchase the asset. Yet they don't need that, they predominantly need the housing services and tenure benefits that homeownership provides.
Point is, how would we know if rents would rise, we don't know. We also don't know what changing property taxes will do with any certainty. We do know that the price of housing has gone stratospheric and that can't be based on any input costs. I know you have published some comparisons of different jurisdictions but it's difficult to draw any firm conclusions about these relationships from purely econometric data. We need a model that takes everything into account and fits the data in multiple jurisdictions. Your calculation is extreme hard to interpret in isolation.
Based on the assumption that house prices are determined by the marginal borrower, removing stamp duty would just push the price up because borrowed funds would be diverted from the tax to bidding up house prices.
How sure are we that house prices are determined by "yield"? What if prices are pushed up by financialisation, speculation and bank lending, and rents rise as a result?