Then again, the deduction probably encourages charitable giving that wouldn't happen otherwise. For example, the CBO found that limiting the deduction to giving above 2 percent of one's income would reduce charitable giving by $3 billion annually. Not all reforms have that effect — setting a floor but making the deduction available to non-itemizers raises money without hurting charities — but many do.
This suggests that whatever savings these reforms produce comes at the expense of charities. The Tax Policy Center, for example, found that replacing the deduction with a refundable credit would reduce giving by between $6 billion and $10.8 billion, and would increase federal revenue by $9.4 billion to $10.4 billion. There's a small possibility, in other words, that that proposal would cost charities more than it earned the government.
But that's only a problem if the giving being subsidized is actually worth the subsidy. Evaluating that is a value judgment, of course, but it's worth noting that the vast majority of charitable giving doesn't go to things like homeless shelters and soup kitchens. According to the Giving USA report, which tracks charitable giving patterns, the biggest category of recipient is religious organizations. Note that that just includes donations to churches, synagogues, mosques, etc, and not donations to religiously inspired service groups like the Salvation Army or Episcopal Relief and Development.
Next up is education — including donations to universities and private secondary schools — and only then do you get to human services, where groups like the Red Cross fall. So what share of charitable giving actually goes to poor people? Unfortunately, we don't have up-to-date data on that, but in 2007 Google and Indiana University's Center for Philanthropy tried to find out the answer for 2005's donations. That's not as recent as one would like, but it does give a sense of what the breakdown was prior to the financial crisis, which really battered the nonprofit sector.
The assumptions of the report are quite generous to givers. For example, when faced with donations that partially benefited poor people but also helped the non-poor (for example, donations to hotlines that get calls from people across the income spectrum), they counted half the donations as donations to the poor. But even so, the report found that only about a third of donations in 2005 were targeted at helping the poor:
Particularly startling is that even when you lump in international aid to the poor (and, of course, the poor in developing countries are much worse off than the poor in developed country) with all miscellaneous donations to help the poor, it's still less than 5 percent of all donations.
Now, maybe helping the poor isn't the only, or even the main purpose of the charitable deduction. Maybe part of being a liberal society that tolerates a variety of religious traditions is giving religious people tax breaks on what they give to their churches. Maybe donations to higher educational institutions, even donations that are poorly disguised bribes to get the donors' children admitted, are worth subsidizing.
But regardless of its other merits, the deduction's distributive argument seems pretty clearly regressive.