I've never thought of myself as a stupid man, particularly when it comes to thinking about numbers and algorithms. In fact at one time (admittedly nearly thirty years ago) I was good enough to be graduated summa cum laude with a degree in computer science and applied mathematics. Which is why I find myself so bothered by the explanation and discussion of the so-called "chained CPI".
The objection I see over and over again to applying "chained CPI" to Social Security is that it doesn't accurately reflect the kinds of cost increases that affect seniors:
Some analysts fear the chained CPI switch unfairly burdens seniors, who tend to have more health care needs–which, unlike food or clothing choices, are not “elastic.” If you need blood-pressure medication, you still need it even if (or when) the price rises. You can’t simply choose to have a different medical condition that happens to be cheaper to treat. And if you’re not free to substitute chicken for steak, the fundamental economic assumption that drives the chained CPI no longer applies. “Seniors have a different consumption basket from the young, one that includes more medical expenses, and [they] probably face true inflation that’s higher, not lower, than the official measure,” wrote Paul Krugman in The New York Times Friday. “This is, purely and simply, a benefit cut.”
and that
CPI-E is a better measure.
I'm sure that's true. But my difficulty is far more fundamental. It has to do with the reasoning behind chained CPI, which I'm afraid I can't make sense of no matter how hard I try. Here's a typical explanation:
The chained CPI slows the growth of entitlement programs by assuming that when prices go up for something, people switch to cheaper substitutes. That is, if the price of steak goes up, people will skip the steak and buy chicken instead. And if people aren’t necessarily paying more for their goods, they don’t need their benefits to rise. Estimates show that under the chained CPI, your cost-of-living adjustment would be about .3 percentage points below what it would be under the plain CPI, which means you won’t see as big an increase in your Social Security check.
Here's basically the same from another source:
How it works: The new measure assumes that people change their buying habits when things become more expensive, so they don’t need their benefits to rise as much every year. If they’re buying groceries, for example, and the price of steak is rising, they could buy chicken or fish instead.
The problem I have is this: this is an iterative process. This lower rate of increase is applied each year, on the assumption that each year people are making these kinds of substitutions of cheaper alternatives.
But there is not an infinite supply of cheaper alternatives. Once I have substituted chicken for steak, I can't do that again next year, and the year after that, and the year after that. At some point, one runs out of cheaper alternatives. If I'm living on generic dry cat food and tap water, what's my substitution going to be next year?
To make matters worse, anyone who is living at the edge of subsistence will already have made almost all available substitutions. So they hit this wall in the very first year.
This is such an elementary problem with this whole idea that I can't help thinking I must just be misunderstanding the whole concept. If there really is such a basic error in the whole idea, surely more people would be talking about it. Maybe these popular explanations are just flat-out wrong. Or maybe early-onset dementia is hitting me at age 52 and I've just lost my knack of applying algorithms to the real world.
Whatever it is, I just don't get it.