Chaining ourselves to (slightly) higher tax rates

I was going to suggest a couple of potentially good things about chained-CPI, a slower and allegedly more accurate measure of inflation than the one currently in use, as a means of boosting tax revenue. Josh Barro slowed me up. Having criticized chained-CPI as a means of reducing Social Security benefits, which Barro believes should be indexed to income growth rather than inflation, he moves on to taxes:

Inflation indexing of the income tax code also makes little sense. Every year, income tax brackets are adjusted upward in line with CPI. So, while the 25 percent federal income tax bracket started at $34,500 of taxable income in 2011, it doesn't start until $35,350 for 2012. But, except in recessions, incomes tend to rise faster than price inflation. That means that, absent changes in tax law, a taxpayer at any given place in the income distribution will face a higher effective tax rate over time.

This effect is called "real bracket creep," and it’s undesirable if we want a tax code that produces stable collections and a stable distribution of the tax burden over time. Indexing tax brackets to national income would cause the bracket thresholds to rise faster, eliminating real bracket creep.

If inflation indexing for brackets is bad, because it adjusts brackets too slowly, chained-CPI is worse.

The thing is, Barro's arguments are premised upon the possibility of rational policymaking, in which enough tax revenue would be collected to meet the country's needs without subterfuge. You can only forgo letting rates ratchet up by a too-fast automatic mechanism if you set them at an adequate level in the first place. That is not going to happen in the United States, with the GOP in the eternal grip of antitax fanaticism.  Hence, as Andrew Sullivan says about chained-CPI for both benefits and tax levels:

I think it's one flawed way to reconcile Americans' unreasonable demands of government with their equally unreasonable refusal to pay for it.

With that extended caveat, on to the couple of advantages of using a chained-CPI for tax purposes. First, I have always thought that Obama's promise to raise taxes only on those earning more than $250,000 ridiculous. Acknowledging that many Americans are hurting, and that incomes have long been stagnant for earners in the lower four quintiles, and that the tax system should be made more progressive to offset rising income inequality, we still need more national income than can be efficiently extracted from the top 2%; we need to inch our way back to at least Clinton-era tax levels (or eventually, a revamped equivalent, relying on some mix of energy and consumption taxes).  Chained-CPI starts that crawl-back -- modestly, adding all of about $60-90 billion in new revenue over ten years.

Second, as long as inflation is super-low, chained-CPI will raise taxes very gradually. And at present, as long as economic growth remains tepid, inflation will remain low.  Thus, chained-CPI will only start to bite incomes significantly as the economy heats up, which is as it should be. I once suggested, pretty much in jest, that marginal tax rates should adjust automatically according to GDP growth.  Chained-CPI offers a very faint equivalent -- at least until we run into another round of stagflation.

To some extent, I'm granting here the Ezra Klein premise I took issue with yesterday: that chained-CPI is a stealthy and hence dishonest method of cutting benefits and raising taxes. But I didn't mean to argue that it's not stealthy -- just that it's not particularly dishonest, since it puts our calculation of benefits and tax rates on a more accurate statistical model (if Klein is right to grant that premise, as he does). Per Sullivan above, we need stealth.