The transfer price in gp’s example would be zero, wouldn’t it.
What probably happens is that I sell you my share in the holding entity for $150k, and the entity sells me the house for $150k.
There are cases where you could simply dissolve the entity and distribute the assets, which would avoid the sale and resulting taxes, in which case you’re just re-titling it and not selling it.
Tl;dr: this gets complicated really quickly.
It seems like the question here is how long you can delay taxes on a capital gain. It doesn't seem particularly unfair if you can't always delay capital gains tax as long as you like?
Company starts with $200K, held 50:50 by two shareholders, losslessly buys two house for that, those go to $300K, company losslessly sells one of them for $150K, company pays $20K tax on $50K gain, and then owns a $150K house and $130K in cash. Shareholder B sells their 50% share in the company to shareholder A for $140K cash, realizing a $40K capital gain on their original shares investment [incurring $16K in taxes on the gain in their shares]. Shareholder A now owns all the shares with a basis of $240K for a company value of $280K.
I argue that Shareholder B’s economic interest in a single gain has been taxed twice.
Later, when/if I sell the house, tax is due on the $50K in gains. Just like in the ETF case.
In the alternate treatment seemingly contemplated by the article, we could instead agree to sell one of the houses, jointly pay taxes on $50K in gains, and give me the sum of $150K-taxes, the gains of which I'd pay taxes upon again. You'd also have effectively paid taxes when you hadn't done anything but invest and hold. There's a pretty good argument this is unfair to me (to tax me twice on one gain); there's an almost ironclad argument that it's unfair to you.
It's no surprise that the treatment that is both more reasonable and more profitable is the one that ETFs seek to implement.
(Fees and other weirdnesses around real-estate ignored above for simplicity.)