[tim333]

Maybe but say YC put $20k into Airbnb for a stake now worth $1bn and $80k into 4 other statups that went bust. GAAP would show a net loss of $80k which would not reflect very accurately how they were doing. GAAP has been designed more to provide a consistent way of calculating corporate income tax than to accurately show how a company is doing.

[lugg]

You're welcome to suggest a more accurate way of consistently accounting income tax requirements.

[hn_throwaway_99]

That seems like an ungenerous interpretation of the parent's comment. I didn't take that comment to mean that GAAP is not a good way to calculate income tax, but that it's not a good way to judge the true financial prospects of a company.

[lugg]

I took it the same way. If there is a better way to judge the true financial prospects of a company we should be taxing them accordingly.

My point was that GAAP is probably the best we're going to do because of the consistency problems.

The moment you start applying special rules or caveats per company you stop being able to compare companies properly as an investor, at which point the measure becomes useless as a talking point as well as being useless for tax calculations.

My original post was framed against tax because it doesn't really make a difference. If the new method is good enough for comparison between companies it's good enough to replace GAAP - unless it's not.

[hn_throwaway_99]

I don't agree with this sentence at all, "If there is a better way to judge the true financial prospects of a company we should be taxing them accordingly."

The main thing investors want are some measures that are reasonably forward looking, with the caveat that some of those forward-looking measures will of course be inaccurate when the future comes.

Taxes, on the other hand, should be mostly past looking, taking into account events that are fully settled and guaranteed.

If, as an investor, you're only examining trailing metrics, you're going to be a pretty horrible investor.

[tim333]

It's actually an interesting question as to whether something different from GAAP should be used for tax. For companies based in one country it's a time honoured system that has worked for decades but for multinationals it's become in my opinion too easy for them to use loopholes and pay little or no tax.

Like with YC they are US based and if they sell their dropbox stake they'll be taxed but more tax avoiding companies might say have held the dropbox stake through a Cayman Islands trust or some such and try avoiding it. I'd vote for something like for multinationals they do accounts as if all their multinational bits were in one country and then allocate the earnings to different countries in proportion to the sales there. So if Microsoft had a lot of Windows sales in Germany say they'd pay tax on the profits there rather than saying they were all generated at a DVD stamping plant in Ireland for example.

[beambot]

Easy: they only realize taxable gains when the startups have a liquidity event -- ie go public or get acquired. (Or die, for the losses.)