More at Stake With the Chained CPI Than Social Security Benefits

January 25, 2013

As perhaps you know, the President’s last comprehensive “fiscal cliff” proposal included the adoption of a new inflation measure — the Chained Consumer Price Index.

We’ve heard about it before, but almost exclusively as a way to curb spending on Social Security retirement benefits. And that’s what we’re hearing most about now.

But what the President proposed was apparently a global switch to the chained CPI, with some unspecified protections for “the most vulnerable.”

Just because Congress decided to punt on the “fiscal cliff” doesn’t mean something of this sort won’t resurface.

In fact, it already has in a bill co-sponsored by Tennessee’s Republican Senators Bob Corker and Lamar Alexander.

Like the President’s offer, the Corker-Alexander bill would make the chained CPI the inflation measure used for all federal cost-of-living adjustments.

So it would raise more tax revenues, with the highest percent increases coming from fairly low-income households.

It would also make relatively fewer people eligible for a host of safety net programs and, in some cases, reduce the benefits those still eligible would get, relative to what they could expect if there were no CPI switch.

The same result, of course, for Social Security retirement benefits.

I’ll deal here with the eligibility issue and return to benefit cuts in a separate post. But first, a super-simple primer to set the context.

Chained CPI 101

At this point, the federal government uses the Consumer Price Index for All Urban Consumers (CPI-U) to make what are basically cost-of-living adjustments in both the tax code and the Census Bureau’s poverty thresholds.

A somewhat different index — the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W) — is used to adjust all major federal retirement benefits, including Social Security, and Social Security disability benefits as well.

Both these indexes reflect the prices consumers pay for a set market basket of goods and services. When the average costs of the total go up, so do  the CPIs.

But, say economists, consumers change their purchasing practices when the costs of certain items rise. For example, if the price of beef goes up, they buy less of it and more chicken.

The chained CPI is designed to capture these changes. So the cost-of-living increases it produces are lower than those based on a a market basket that isn’t continuously re-weighted to reflect substitutions.

Eligibility for Safety Net Benefits

More than 30 federal anti-poverty programs based their income cut-offs or targeting on the federal poverty guidelines,* as do some state and local programs.

The federal programs include many we think of as key parts of the safety net, e.g., the food stamp program, other nutrition assistance programs, the Children’s Health Insurance Program, LIHEAP (the Low Income Home Energy Assistance Program).

The poverty guidelines are simplified versions of the Census Bureau’s official poverty thresholds. These, as I’ve said, are annually adjusted using the CPI-U.

The chained CPI would thus mean smaller upward adjustments in the income cut-offs — not much smaller in any given year, but cumulative over time.

Bad Policymaking

I see at least three problems with the result I’ve just described.

First, the thresholds already significantly understate the number of poor people in this country, as even the Census Bureau’s still-evolving Supplemental Poverty Measure shows.

This is partly because the thresholds are based on an outdated minimum cost-of-living measure — three times the cost of what used to be the U.S. Department of Agriculture’s cheapest food plan.

The annual adjustments compound the problem because the CPI-U understates living-cost increases for households in the bottom fifth of the income scale — or so the research we have suggests.

Use of the chained CPI would thus, as Shawn Fremstad at the Center for Economic and Policy Research says, define deprivation downward, even more than use of the current thresholds do.

At the same time, the living-cost research tells us that the chained CPI probably isn’t more accurate for low-income households — quite the opposite, in fact.

So using it would unjustly exclude even more people from the safety net — assuming the proper measure is insufficient income to pay for basic living needs. And if not that, what?

Finally, the proposed switch to the chained CPI is an underhanded way to make consequential policy changes. And that, as Wonkblogger Dylan Matthews says, is what makes it so attractive.

If our policymakers want to shrink the safety net, then they should say so forthrightly, name the programs and give us the figures — not trot out a supposedly technical change that only the most wonkish among us can understand.

* The official list of programs that use the federal poverty guidelines says that the Temporary Assistance for Needy Families Program doesn’t. This is because states can set income eligibility standards for TANF however they choose. Fourteen expressly use the federal poverty level, which I assume means the guidelines.


New Reasons New Census Figures Won’t Give Us a True Read on Poverty in America

September 6, 2012

Next week, the Census Bureau will issue the first of its annual reports. Economists surveyed by the Associated Press predict that the poverty rate will rise again.

Two recent blog posts tell us that whatever the Bureau reports next week — or later, when it issues the results of its more detailed American Community Survey and its supplemental poverty measure — will understate the number of poor people in America.

Another Problem With the Poverty Thresholds

The poverty rate we generally read about is based on a set of thresholds the Census Bureau uses.

The thresholds matter not only because they’re our main source of poverty data, but because they’re the basis for the federal poverty guidelines. They thus ultimately determine eligibility for a wide range of safety net programs.

It’s common knowledge that the thresholds are a crude, out-dated poverty measure — three times what the cheapest U.S. Department of Agriculture food plan cost a family of four in the mid-1960s, adjusted annually for inflation.

Blogger Evan Soltas — a super-wonkish undergraduate, even by Princeton standards — adds a new wrinkle. The inflation adjustments themselves, he says, cause the Census Bureau to under-count the poor.

The Bureau uses the CPI-U (Consumer Price Index for All Urban Consumers) to adjust its thresholds. The CPI-U reflects the cost of a market basket of goods and services commonly purchased by people who live in metropolitan areas.

But, says Soltas, the market basket of goods and services purchased by people in the bottom fifth of the income scale is different. And its costs have risen significantly more than the CPI-U.

The cumulative difference since 1967 is a 12% understatement of living costs for the poor. Hence an under-count built on top of the under-count resulting from a drop in food costs as a percent of total household budgets.

Any safety net program that indexes to the CPI-U has thus effectively cut benefits by the same 12%, Soltas says. I assume this includes all the programs that use the poverty guidelines.

More than 30 federal programs do. Some state programs also. Seems we’ve got a big problem then — different from the big problems we already knew.

A Problem With the Supplemental Measure Too?

The Census Bureau’s supplemental poverty measure takes an altogether different approach to the thresholds, basing them on the 33rd percentile of what households with two children spend on four basic needs — food, clothing, shelter and utilities — plus a multiplier to accommodate the rest.

It also factors in some other “nondiscretionary expenses,” on the one hand, and major federal benefits that don’t come to recipients as cold cash, e.g., tax credits, food stamps.

The results are commonly viewed as a big improvement over the official Census figures. Shawn Fremstad at the Center for Economic and Policy Research says not necessarily.

In 2010, the child poverty rate was 4.3% lower under the SPM than the official measure.

Conversely, the senior poverty rate was 6.9% higher — mainly because the SPM takes account of out-of-pocket health care expenses.

But look, says Fremstad, at USDA’s food insecurity rate — the “most established measure” we have of “direct deprivation.”

According to Fremstad, the 2010 food insecurity rate for children was 20.2% — closer to the official poverty rate than the SPM rate. An even smaller difference between the food insecurity and official poverty rates for seniors.

Fremstad thinks the SPM rate for seniors is about right. The child poverty rate isn’t because it fails to capture the unique costs of meeting children’s “basic needs for care and healthy development.”

No one, I think, could argue with that. Whether the food insecurity rate Fremstad cites reflects “direct deprivation” experienced by children is another matter.

We know a family can be food insecure even if it always has the resources to buy enough food to keep children from going hungry. Adults will skip or scrimp on meals first, as the USDA data clearly show.

Fremstad is comparing the child poverty rates with the food insecurity rate for households with children. The food insecurity rate specifically for children is just under half that.

So it’s not food that most officially poor children are missing, though we’ve disturbing reports of children showing up at school hungry.

It’s those other investments in their healthy development — the parental attention, high-quality child care and other resource-based influences that account for wide income disparities in school readiness among kindergarteners and subsequent academic performance.

Hard to imagine the Census Bureau could measure these. But Fremstad seems to think it should try. Because it will otherwise “continue to define deprivation down for America’s children.”