Random policy idea: family price indices

And I think it would work, too!

We should produce monthly price indices to track how price changes are affecting different households—high or low income, older or young, with or without children, with or without very young children, in urban and rural areas.

The Consumer Price Index (CPI) is produced and released monthly. The headline reads something like: “Prices rose 0.1 percent last month, and 4.0 percent from the year prior.”

This overall number, which is sometimes called the “all items index,” is accompanied by additional enumerated indices by type of item, like the shelter index, used cars and trucks index, energy index, food index, and so on.

I’m proposing: Have additional enumerated indices produced monthly by type of consumer, like:

• total income of the household, regardless of the composition: richest 10 percent, poorest 10 percent, middle 30 percent

• family type: single adult living alone, couple without children, one-parent family, multigenerational family,

• child-family type: age- number combination of children,

• spending burdened: households that spend much higher of a share of income on average on big necesseties, like rent, energy, food, or medical care.

In total this would means dozens of new indices, but we’d also have a heat map of which Americans could be struggling to afford goods. Infant formula recalls affect a different set of people than airfare hikes, same with egg shortages and hotels. And of the people affected, they are affected differently based on their income. I even think there could be subindices where we track necessities versus luxuries, within each group. A high-income family with two children versus a middle- or low-income family with two children might have similar necessities but different luxuries.

Can we do this?

The pieces are there, we just need to put them together. And in a way, we already do.

CPI is produced by the Bureau of Labor Statistics (BLS). To produce the index, the “I” of CPI, they do three things (not necessarily in this order). First, they go out and collect prices from sellers. Second, they decide what prices to include in the index. This is called the “market basket” and the prices tracked in the index are the prices in the defined basket. Third, they weight prices in the market basket based on how important they are to spending, and the result is the index. Weighting means that not all prices are equally influential within the index; something like rent and food is large, but watches are small. This is also why it’s called an index and not an average, because it’s not a true average.

To do these three things, the BLS is combining data from two places. The data on prices they collect from tens of thousands of retail stores and landlords; this is the data published in the CPI. The data on spending, on the other hand, is from the Consumer Expenditure Survey (CEX). The Consumer Expenditure Survey is also produced by the BLS, and in it, respondents log what they actually spend money on, everything from taxes to food at restaurants. The spending data in the CEX is used to define which items are in the CPI market basket and how much it item should be weighted.

The market basket is not defined point-in-time, but with an intentional lag. Typically the basket in one year (say 2021) is a reflection of average spending over multiple years, from multiple years prior (2017-2018). The CPI is not meant to capture short-term changes in spending, but short-term changes in prices, so the market basket is designed to be stable. Yet, spending habits and patterns do change over time, so the market basket is designed to change with long-term trends.

CPI gives us how much stuff costs and CEX gives us what people buy, and this data is already combined to define the market basket.

To make family-, household-, or income-specific price indices, you’d do the same thing you do for the market basket, but for subgroups of the population. With smaller groups, it’s possible you’d have to use more years of data to find the most consistent basket and weights, but it’s not impossible.

It’s worth noting that CEX has many subgroups that it tracks already, which it calls “consumer units,” based on income, geography, and family type. It publishes annual expenditure information on them as well.

And that’s it. Make more price indices that reflect what we know about people so we have a more accurate sense of who is being affected by, or driving, inflation. Like I said in the subhead, I think this would work.

For a historical perspective, keep reading.

There’s some parallels of this proposal to the work of Mollie Orshansky. She was a long time government economist and was at the frontier of researching family budgets, necessities, and costs of living. At the Department of Agriculture after World War II, she and a colleague answered letters from the public about how to make their income meet food budgets, and what food they should prioritize. In this role, she was akin to a public interpreter of the USDA’s food plans. These are exactly how they sound: they are meal plans to achieve a healthy diet. There are four tiers of food plans based on the amount of money a family can spend: thrifty, low, moderate, liberal. Orshansky would write to families helping them plan meals on a budget.

Later, she worked with a survey the USDA fielded to understand how much food accounted for overall family budgets (the 1955 Household Food Consumption Survey) and found that families spent about a third of their budget on food. She moved to the Social Security Administration and researched family budgets, estimating the necessary costs for living from various aspects or for various family types.

In the days before Social Security benefits were automatically adjusted every year for inflation, benefits would stay fixed in nominal terms until Congress raised them. There’s nothing to say that they were any good at this. Had they raised them enough or raised them too much? Certainly it’s a political win for members of Congress to go home and say they raised your Social Security benefit, especially in an election year. But it wasn’t clear if they were hitting the mark on benefits, because the mark hadn’t been established. There was not an accepted understanding of how much it took to get by in retirement, or a methodology for how to estimate it. This was part of Orshansky’s job, putting numbers to the abstract idea of “having enough to get by.”

As part of her work for Social Security, Orshansky was assigned a project to estimate how poverty affects children. This was before the federal government had any consistent or established poverty method. To understand how poverty affects children, she first had to come up with a way to determining which children were in poverty. The USDA publishes how much food costs in its food plans and consumers reported in 1955 how much money they spent on food; if you link the two, you can come up with a necessity threshold, even if you only know the price of one category of spending. Being in poverty is having less than the cheapest food plan multiplied by three. It’s a very clever use of data in that its resourceful, straightforward, and reasonably accurate.

The story breaks here. One branch heads to public policy lore: Orshansky published her findings of children and poverty thresholds in two articles in the Social Security Bulletin in 1963 and 1965, and they appeared right as the Johnson administration declared a war on poverty. Three times the cost of food became the accepted poverty threshold, but rather than increase the threshold every year to account for 1) change in food prices or 2) change in spending, the 1959 threshold was simply indexed for inflation. Orshansky had developed the official poverty measure. Or more accurately, a version of a measure she developed was then used as the official one. When she died, she was eulogized as Ms Poverty, though she had always been clear that the measure was not meant to be used the way it was. Her legacy has become more about her work to count the poor, rather than the measure the government adopted. Regardless, her work in the 1960s is still used today to determine whether tens of millions of Americans are eligible for social welfare benefits, from SNAP to TANF to housing to SSI.

But go back to 1963, before the White House finds her work, simplifies it, and makes a poverty measure from it, and follow the different branch. Orshansky is diligent. She has spent two decades researching how families afford basic necessities. She is tasked with finding out the number of children in poverty, determining some threshold to apply. Her original paper didn’t have a single threshold, but 124 of them. As her Social Security remembrance explains:

“She differentiated her thresholds not only by family size but also by farm/nonfarm status, by the gender of the family head, by the number of family members who were children, and (for one- and two-person units only) by aged/non-aged status.”

She did this twice, 124 thresholds according to the thrifty (aka economy) food plan and 124 thresholds for the low cost food plan, which she favored. The thrifty thresholds were collapsed into a weighted average that became the poverty threshold, but her work was detailed, speaking to the diverse needs of families. There’s nothing about poverty that says you have to default to an average, and pioneering poverty researchers like Orshansky looked at far more detail.

And if Orshansky can do this in 1963 by hand to come up with poverty, I feel confident that we can produce monthly family price indices in 2023, and I’d be thrilled to see 248 of them. As she noted in a retrospective in 1988 for the Social Security Bulletin about the poverty measure:

“No information is available on how families at different income levels adjust spending to accommodate changes in prices, or how higher expenditures in one category—such as health care—affect other aspects of family living. Yesterday’s luxuries quickly become tomorrow’s necessities…”