Kevin Drum writes against the tax preference for capital gains income, and links to an informative discussion (PDF) from the Congressional Research Service. One point the CRS notes, but really should have been clearer about, is that when you're addressing this question you need to distinguish between an offset and a non-offset scenario. When you talk about a debt-financed cut in the capital gains tax rate, in order to increase the overall level of savings and investment in the United States you need to posit a huge behavioral response to the tax change. When you look at the recent history of the capital gains tax rate, this is mostly what we're talking about.
The theoretical argument that lower capital gains taxes should spur savings and investment is much more compellingly applied to a balanced budget scenario. But in this case we'd have to examine what the offsets are. If you reduce taxes on investment income and finance that by reducing public investment in infrastructure and education, then you'll consider that a growth-boosting tradeoff if and only if you think the public sector massively oversupplies us with infrastructure and education goods. Alternatively, if you reduce taxes on investment income and finance that by cutting Social Security benefits, you'd almost certainly increase savings and investment but you'd be thrown out of office for promoting an insanely inhumane giveaway to the rich. The politically and theoretically sound alternative is to offset your cut in taxation of investment income with new taxes on either high-end consumption or environmental degradation or both. It's striking, however, that the political entrepreneurs promoting low levels of taxation of investment income never propose these options and seem generally unconcerned with the entire question of offsets. It's enough to make you question their sincerity! A path to an increased national savings rate over the long-term that opens with a gigantic increase in public sector debt doesn't make any sense, and the fact that this is the strategy they keep promoting is surely on the list of reasons that their claims are so hard to verify in the data.
But to be smart and sophisticated about this when arguing at the dinner table, insist on making the distinction. If we reduce taxes on investment income and increase taxes on fancy horses, the claim that the Romney family will shift money away from horses and into investments makes perfect sense. If we borrow a bunch of money in order to reduce taxes on investment income, then the behavior response of the Romney family is unclear and the net effect on the national savings rate could easily be negative. What's more, the horse tax option doesn't have regressive distributional consequences but the borrow-the-money option does. If you see the regressive distributional consequences as a feature rather than a bug, then of course borrow-the-money may look appealing, but if not there's no reason to embrace it.