Marco Rubio says he’d cut taxes on the poor more than the rich. Nonsense.

Marco Rubio insisted to CNBC’s John Harwood at last week’s Republican presidential debate that he’d cut taxes more on the poor than on the rich. That wasn’t true given the information he had at the time, and it’s definitely not true if you look at a new analysis conducted by Citizens for Tax Justice, a left-leaning think tank.

These are the first estimates of his plan’s impact from a group other than the Tax Foundation, the right-leaning think tank he leaned on for his campaign’s own analysis. And they show the top 1 percent making out like bandits:

This analysis is probably overly generous to Rubio. Like the Tax Foundation, Citizens for Tax Justice assumed that the new $2,000 for singles, $4,000 for couples tax credit that Rubio is proposing giving to every household is fully refundable. Rubio’s team has insisted to me that that’s not the case, and that they’ll institute reforms to prevent people without any income from getting thousands of dollars from the government, no strings attached. “Rules would be tailored to ensure that our reforms would not create payments for new, non-working filers,” one of Rubio’s aides told me.

For that reason, CTJ’s president, Bob McIntyre, told me the estimate for the lowest 20 percent should “come with an asterisk on it.”

CTJ has also analyzed Donald Trump and Jeb Bush’s tax plans, and Rubio’s proposal cuts taxes for the top 1 percent about as much as Trump and more than Bush. Rubio would totally eliminate all taxes on investment income, which goes overwhelmingly to rich people:

Rubio’s plan costs more than Bush’s plan as well. While CTJ estimates that Bush’s tax plan would cost $7.1 trillion over 10 years, it finds that Rubio’s would cost $11.8 trillion (Trump’s plan costs about the same as Rubio’s, according to CTJ’s most recent numbers). CTJ’s estimates for Trump match up with the Tax Foundation’s, but that group models Bush and Rubio’s plans as much cheaper. “For Bush and Rubio, the large differences seem to be mostly about measuring the effects of the corporate tax cuts, although I can’t say for sure,” McIntyre says. “Obviously, I think that we’re right, and that the Tax Foundation vastly understates the cost of the corporate tax cuts that Bush and Rubio have proposed.”

It’s true that under CTJ’s analysis, the bottom 20 percent would get an average tax cut worth 13.9 percent of their income, while the top 1 percent would get one worth 12.5 percent of their income. But that’s a misleading comparison. For one thing, Rubio’s plan probably wouldn’t help the poor that much once restrictions on its new credit are imposed. More importantly, percentages are a bad way to do this kind of comparison.

According to the above analysis, the top 1 percent would, on average, get tax cuts more than 103 times larger than the poorest 20 percent would get. And that’s before considering that the benefit for the bottom 20 percent is exaggerated. Rubio insisted at the debate that this doesn’t matter: “Yeah, someone who makes more money, numerically, it’s gonna be higher. But the greatest gains, percentage-wise, for people, are gonna be at the lower end of our plan.” But who gets more money “numerically” is actually what matters here. Tax cuts cost money. That money can either go to poor people, or it can go to rich people. And under Rubio’s plan, much, much more of it would go to rich people.

Footnote: McIntyre explains that the Bush tax plan numbers estimate changes in 2015, while the other two plans’ estimates are for 2016. That’s just a function of the Bush plan estimates being prepared earlier, and doesn’t make a real difference in the distributional analyses.

Update: After publication, McIntyre submitted updated estimates for Trump’s plan showing an even bigger benefit for the top than in Rubio’s plan. The post has been updated accordingly.