The devastating effects that many provisions of the Republican Senate and House tax bills would have on the middle class have been well documented, thanks to the efforts of independent analysts at the Tax Policy Center and elsewhere.
But here’s a tax increase on working Americans that hasn’t been very well publicized, even though it’s in both GOP proposals. It’s a change in the inflation measure on which tax rates are calculated to something called the chained consumer price index.
I’ve written about the chained CPI many times in the past, generally when it’s been put forth as a new way to calculate cost-of-living adjustments for Social Security. In that context, it’s a stealth benefit cut. Up to now, it has always been rejected — most recently in 2013, when President Obama tried offering it to congressional Republicans as a concession on the budget, and had to get educated about its downside by his fellow Democrats.
In their tax bills, Republicans in both chambers are proposing to convert the index used to make inflation adjustments to tax brackets and other inflation-sensitive provisions of the tax code from the standard CPI to the chained version. The argument is that it’s a “better” or “more accurate” inflation gauge than the standard CPI.
But that’s just blowing smoke. What Republicans really like about the chained CPI is that it tends to show a lower inflation rate than the alternative, typically by 0.3 percentage point a year. The difference pencils out to significant dollars: Changing the inflation index immediately would raise about $125 billion over the next decade and nearly $500 billion in the decade after that, according to the Tax Policy Center.
Most of that money would come out of the pockets of middle- and working-class taxpayers. Most important, it would slow inflation adjustments to tax brackets. This would hurt those taxpayers because more of them would move into higher tax brackets purely because of inflation in their wages. High-income taxpayers who already pay the top tax rate wouldn’t be affected the same way.
The chained CPI also would reduce inflation increases in provisions such as the standard deduction, personal exemption, earned income tax credit and the alternative minimum tax. Personal exemptions and the AMT are marked for repeal in the GOP tax bills, but if they survive or are supplanted by other inflation-indexed provisions, the chained CPI would represent a tax increase on their users. In any event, as Howard Gleckman of the Tax Policy Center recently observed, the Republican plan is to replace the inflation-indexed personal exemption with a larger but non-indexed child tax credit. Leaving the latter unindexed would erode its value over time.
What’s insidious about the chained CPI is that its effect is modest at first, but compounds over time. After 10 years the shortfall compared with the standard CPI is more than three percentage points; after 20 years, more than six. “Unlike proposals to repeal, say, itemized deductions,” Gleckman writes, “people might not notice this hidden tax increase for years to come.”
One other thing: Inserting the chained CPI into the tax code could be a Trojan horse to infiltrate other government programs, notably Social Security. But that would be only the start. The Congressional Budget Office noted in a 2010 report that the federal poverty level is indexed to a standard CPI; changing that to the chained CPI would affect all programs whose budgets or benefits are keyed to the poverty level, including Head Start, food stamps and “the National School Lunch Program, the Low-Income Home Energy Assistance Program, the Children’s Health Insurance Program, and parts of Medicaid.” Republicans already have their daggers out for some of these programs; the chained CPI would represent just the first prick.
None of this would be as important if the chained CPI really could be justified as a “better” or “more accurate” inflation index than the conventional CPI. But it’s not.
To begin with, no inflation indices are “accurate” in terms of measuring that elusive quality known as inflation; they can only measure price changes within the market baskets they’re based on. If some commodity, product or service isn’t part of that basket, then its price changes aren’t counted. The best that index designers can do is assemble a market basket that resembles the monthly purchases of a target consumer group — typically urban wage-earners and consumers — and hope they get it close.
One troubling factor recognized by economists is that when prices rise for some item, consumers buy less of it, stop buying it or buy something else instead. This so-called substitution effect makes it hard to compare consumer costs in one month to those in the next. Conventional CPIs try to address that by adjusting the market baskets every two years, but that’s a crude process.
Enter the chained CPI, which the Bureau of Labor Statistics started publishing in 2002. The chained CPI tries to track substitutions month to month, rather than over years, and uses higher-level substitutions. The traditional CPI anticipates that if Gala apples rise in price, consumers might switch to Fuji apples if they’ve risen less; the chained CPI posits that if apples rise in price, consumers might switch to bananas — or more important, if the price of food goes up, they may cut back on heating oil. Even if substitution might reduce one’s standard of living, it does tend to reduce household expenditures, so by that benchmark an index that doesn’t incorporate substitution will overstate inflation.
Even if one grants, for argument’s sake, that the chained CPI is more “accurate,” that would be true only over the long term. Month-to-month, it’s very rough. The BLS publishes a preliminary estimate of the chained CPI about two weeks after the end of every month, at the same time as its traditional CPI release — but revises the chained CPI the following February, and publishes a final figure the February after that. In other words, the chained CPI for any month can be as much as 25 months old.
Some of those revisions can be drastic. For the year that ended December 2009, the Congressional Research Service reported, the initial chained CPI showed an annual inflation rate of 2.7%, the interim revised figure was 3.8%, and the final figure was 2.5%. (The CPI for all urban consumers that year rose by 2.7%.)
In other words, although the CPI has its problems, so does the chained CPI. That’s especially true if it’s used for cost-of-living adjustments in government programs, which may well be the hidden GOP agenda. Replacing the standard CPI with the chained version would mean that “cost-of-living adjustments would either have to wait until the final number was available or rely on preliminary estimates that could change up to two years after the fact,” the CBO noted.
In any event, we know that switching to the chained CPI has nothing to do with accuracy. If Republicans really cared about accuracy, then cost-of-living adjustments for programs such as Social Security wouldn’t use the standard CPI at all — they’d use a measure known as the CPI-E, which the BLS has used to track inflation for the elderly since 1982. The CPI-E overweights spending particularly important to older Americans, such as medical expenses.
But it also exceeds the standard CPI by an average of about 0.2 percentage point a year, so using it to compute the Social Security COLA would mean a larger adjustment for retirees every year. Strangely, one never hears Republican legislators demanding a change to the CPI-E.