Sharing the burden: top houses v top people
People can move. Houses, by and large cannot.
That's one good reason economists would give for removing the 50p tax rate on Britain's highest earners - and taxing its most expensive houses instead.
Another good argument would be that land is taxed relatively lightly in the UK, whereas labour income is taxed a lot.
That's the theory. The reality - economic and political - is, naturally, more complicated.
The Treasury reckons that it will take in £2.8bn more in income tax in the coming fiscal year with a 50p top rate, than if the ceiling were still at 40%.
The Institute for Fiscal Studies believes that could well be an over-estimate. In fact, it has suggested that it might not raise any money at all - because high earners will find ways to avoid it.
This is what the IFS's Mirrless Review of the UK tax system had to say on the matter: "There are numerous ways in which people might reduce their taxable incomes in response to higher tax rates; at some point, increasing tax rates starts to cost money instead of raising it.
"The question is, where is that point?
"Brewer, Saez, and Shephard (2010) addressed precisely this question for the highest-income 1%. Their central estimate is that the taxable income elasticity for this group is 0.46, which implies a revenue-maximizing tax rate on earned income of 56%.
"This in turn (accounting for NICs and indirect taxes) corresponds to an income tax rate of 40%. So, according to these estimates, the introduction of the 50% rate would actually reduce revenue."'Big difference'
In opposition, both David Cameron and the Chancellor George Osborne suggested more or less the same thing.
It's safe to say that a mansion tax on houses worth more than £2m would raise more than nothing. Again, we move house, not the other way around. But how much more it would raise is a matter for lively debate.
At the time of the election, the Liberal Democrats reckoned that a 1% tax on the value of houses above £2m would raise about £1.7bn.
At the time, that struck many as on the high side.
Knight Frank, the estate agency, reckons there are about 45,000 homes in England and Wales worth more than £2m, more than 80% of them in London.
You might be surprised to hear that 40% of them are in Kensington and Chelsea alone. Then again, you might not be.
We don't know whether the average value of those 45,000 homes is a lot more than £2m, or not much more at all. That would make a big difference to overall take.
However, on the World at One on Tuesday, the IFS expert on this, Stuart Adam said £1.7bn was probably "in the right ballpark."
If the 50p rate ends up raising as much as the Treasury first thought, it's looks as though a mansion tax raising £1.5bn to £2bn would not fill the gap.
However, it might be enough to let the government cut the top rate from 50 to 45%.
Those same Treasury estimates suggest that change would only cost the chancellor about £1.1bn next year. If that's true, the Inland Revenue could even come out ahead.
That might sound attractive to George Osborne. But he would then be raising a large amount - from an even smaller number of households.
Raising even £1bn from 45,000 properties means an average extra tax bill, for each of those houses, every year, of more than £22,000.
Britain's top earners may not like paying the 50p rate. But there are a lot more of them - about 275,000, according to the Treasury. (Or at least, that's how many they hoped there would be, after the new top rate had been introduced.)
So, even when it comes to the very wealthiest taxpayers, Mr Osborne may need to decide how much he wants to share the pain.
He may also discover first hand why Britain had traditionally taxed residential property so much less than economists would like - seriously distorting our economy, in the process.
No-one likes chancellors who raise taxes. But British people seem to particularly dislike the chancellors who tax them where they live.
This blog post was first published on Tuesday 6 March 2012. Because of a technical problem, it had to be republished with a new timestamp.