Tuesday, December 18, 2012

At least chained-CPI is chained to reality

The good folks at WonkBlog have been castigating the proposed move to a "chained-CPI" to slow the rate of Social Security benefit growth as "obscurantist,"  as Dylan Matthews called it this morning.  Three hours later, Ezra Klein elaborated the complaint:
Chained-CPI, in [Bowles-Simpson's] telling, is simply an effort to correct a measurement error in the way we calculate inflation. It’s a tweak, a fix, a policy designed to achieve a higher level of technical precision. And who could be against that?

There’s something to this line of argument. The way we measure inflation right now really does mismeasure inflation. Chained-CPI really is a bit more accurate. But that’s not why we’re considering moving to chained-CPI. If all we wanted to do was correct the technical problem, we could make the correction and then compensate the losers.

But no one ever considers that. The only reason we’re considering moving to chained-CPI because it saves money, and it saves money by cutting Social Security benefits and raising taxes, and it’s a much more regressive approach to cutting Social Security benefits and raising taxes than some of the other options on the table.

The question worth asking, then, is if we want to cut Social Security benefits, why are we talking about chained-CPI, rather than some other approach to cutting benefits that’s perhaps more equitable? The answer is that chained-CPI’s role in correcting inflation measurement error is helpful in distracting people from its role in cutting Social Security benefits.
Klein may well be right that there are better ways to cut Social Security than moving to chained-CPI and leaving benefits otherwise unaltered (or that we're better off not cutting Social Security benefits at all).  But there's something fundamentally wrong with his critique, too. If he's right on policy, he's wrong on semantics. And frankly, I'm still trying to figure out whether the semantic error may also imply a policy error.

Klein insists that  chained-CPI "cuts Social Security benefits." That's true, relative to the current baseline. If there's no change, beneficiaries will earn more in five years than they would under chained-CPI.  But if  chained-CPI measures inflation more accurately than the current method, using it would not cut real future benefits compared to what seniors get now. Rather, the current method expands them.  As a method of cutting benefits, chained-CPI has the advantage of being reality-based. 

If the current method of CPI measurement actually grows benefits, there are offsetting facts reducing the current value of Social Security, as Henry Aaron recently enumerated. First, while Social Security payments rise with inflation (however measured), medical costs are rising faster than inflation, and Medicare premiums are deducted from Social Security checks, so retirees' take-home pay is shrinking. Second, the hike in the age at which retirees can claim full social security benefits from 65 to 67 (66 for those born between 1938 and 1954 ,67 for those born 1955 and after) effectively cuts benefits by 13%, at least to the extent that the age at which people actually start claiming benefits does not change. Aaron explains:
Despite the name, that change had next to nothing to do with when people normally retire or claim benefits. What Congress actually did was to raise the age at which unreduced benefits—the amount generated by the Social Security benefit formula—are paid. Congress left unchanged the age when people can and, typically, do first claim benefits, age 62. Retirees who claim benefits before the “full-benefits” age receive less than the full benefit. The reduction is 6.67 percent multiplied by how many years before the full-benefits age that people take their pensions. For that reason, raising the age at which unreduced benefits are paid by two years simply cuts benefits for all retirees by just over 13 percent.
It seems to me, though, that there are arguments against using cost-of-living increases in Social Security to offset these factors. Regarding the Medicare deductions,shouldn't the struggle to bend the healthcare cost curve without cutting seniors' health benefits be waged within the healthcare system?  Is permanently exaggerating the inflation adjustment in Social Security the right offset? I'd rather find ways to cut medical overutilization and cut payments to healthcare providers, since we pay much more per service than any other wealthy country.

Regarding the benefit cut effected by raising the retirement age: oddly, later in the same article, Aaron rather ambivalently advocates raising the eligibility age for Medicare as the least bad form of benefit cut, because it "would nudge some people into remaining economically active a bit longer than they now do," which reduces benefit payouts and boosts tax revenue. Well, doesn't raising the age at which full Social Security benefits become available provide the same "nudge"?  If so, it's not a 13% benefit cut for the nudged -- not if they benefit on the other end by increased life expectancy.

Wonkblog prefers more progressive benefit cuts to Social Security than using chained-CPI, e.g., lowering the percentage of lifetime contributions that high earners get back as benefits while raising the percentage of lower income workers' takeaway.  But that and other progressive tweaks could be done on top of  chained-CPI, with the progressivity made steeper to offset lower income beneficiaries' losses caused by the slower cost-of-living increases.  (The Center for Budget and Policy Priorities recommends that a switch to chained CPI be accompanied by special considerations for the very old and the disabled).

If you grant that chained-CPI is more accurate than the currently used inflation measure, then it seems to me that using it will make for sounder policy in the long run.  


  1. It's funny you brought up the idea of Social Security recipients' checks going down because I was just contemplating the same thing.

    Regardless of how the COLA is calculated, SS beneficiaries' take home pay will not be shrinking for two reasons. First, a 2% COLA increase on a $1,200 benefit is more than a 8-10% increase in a $100 Part B premium. Second, the hold harmless provision of the Social Security Act protects seniors' checks from Part B increases. That means that if you receive a check in December the check you receive in January cannot be less as a result of the increase in Part B premiums. (There are exceptions for high earners subject to the Income Related Monthly Adjustment Amount and those whose cash SS benefits were less than the Part B premium.)

    As to the substance of your article, you're correct about Ezra's faulty logic. However, I disagree with Ezra's premise that CPI-W mis-measures inflation. The claim that chained CPI is more accurate is just sleight of hand to obscure the cut in benefits.

    It may be a Rorschach test, but my reading of BLS's description of how chained CPI measures price changes screams "cuts." Here's the relevant passage from the BLS website: "The geometric mean formula [used in various CPI calculations to allow within category substitution], though, does not account for consumer substitution taking place between CPI item categories. For example, pork and beef are two separate CPI item categories. If the price of pork increases while the price of beef does not, consumers might shift away from pork to beef. The C-CPI-U is designed to account for this type of consumer substitution between CPI item categories."

    Embedded in the claim that chained CPI is more accurate is a value judgment that proponents have dressed up as a normative judgment: That substitution does not decrease one's standard of living. Is losing the purchasing power to buy a specific type of a product a decrease in one's standard of living? Is it a cut? Those who claim it's a technical fix are arguing that it isn't actually a cut. Or maybe it is a cut, but not a "real" cut.

    They may have a point. We're not replacing steak with cat food, we're accounting for changes in behavior brought about by changes in relative prices. That said, I see the restriction in the menu of choices as a decrease in one's standard of living, and hence a cut. After all, if chained CPI is more accurate, why does the CBPP recommend measures to combat the compounded effects after many years?

    A "technical fix" I'd like to see is switching to the CPI-E index that measures inflation based on a basket of goods more representative of those consumed by the elderly. (It happens to run a bit higher because old people's basket of goods are concentrated in more inflationary categories.) This fix is more in keeping with SS's legislative intent in that it allows beneficiaries to maintain a stable standard of living after retirement. However, until CPI-E is adopted, I think CPI-W is the more appropriate measure in that it more closely tracks CPI-E.

  2. I thought "special considerations" was exempting the SSI part altogether.