Ed Dolan's Econ Blog

It’s Unanimous: All Indicators Show Inflation is Slowing, Even the Index of Sticky Prices

To no one’s surprise, today’s inflation numbers from the BLS showed that U.S. inflation is slowing according to almost every indicator anyone has thought to report. Seasonally adjusted monthly data for the headline consumer price index, the core CPI and the Cleveland Fed’s trimmed-mean CPI are running well below the Fed’s 2 percent target, as shown in the first of the following charts. The second chart shows that the year-on-year versions of the same indicators are also falling, and are all now running at or below the target.

The latest BLS data are consistent with the price data from the national income accounts that were released a few weeks ago by the Bureau of Economic Analysis. The implicit deflators derived from the national income accounts use different data sets and different methods than the CPI, but they reach the same conclusion. As the following chart shows, the deflators for GDP, personal consumption, and domestic purchases all fell in Q2, and are all running below the Fed’s 2 percent inflation target.

What lies ahead? Economists are right to warn policymakers and market participants not to rely too heavily on volatile monthly data as predictors of future price behavior. However, that warning applies more strongly to some CPI components than to others. Everyone knows that prices for food and energy bounce around a lot, so the core CPI leaves them out. The trimmed mean CPI from the Cleveland Fed takes a different tack, eliminating the 8 percent of prices that increase most each month and the 8 percent that increase least (or fall most), whether they are food, energy, or something else. Why not extend that reasoning even further, and eliminate not just the prices that are volatile in any given month, but those that are chronically volatile?

That is what the Atlanta Fed does with its index of sticky prices. This indicator begins with the observation that the prices of some products change frequently, even minute by minute, while the prices of others remain unchanged for months at a time. The median frequency of price changes for the goods entering into the CPI turns out to be 4.3 months. The Atlanta Fed classifies prices that change more often than that as flexible, and those that change less often as sticky.

In addition to food and energy, the volatile category includes motor vehicles, lodging, clothing, and footwear. Sticky prices include insurance, apartment rents, medical care, and food away from home. (You can find a detailed list here.)

The main reason some prices are sticky is that that the act of changing them is itself costly. Economists often use the term menu costs. The term stems from the idea that a pizza stand is not going to raise the price of its house special from $4.99 to $5.01 each time there is a small change in the price of onions or pepperoni. Those plastic-laminated menu cards cost too much. In practice, there are more important costs of changing prices than those of reprinting menus and price lists. Psychological factors play a big role. Marketers know that price points matter; $5.01 sounds a lot more expensive than $4.99. Loss aversion also plays a role. Customers are moderately happy when you tell them about a price reduction, but they are more strongly unhappy when you tell them about an equivalent price increase. As a result, your best marketing strategy is not to change prices more often than you really have to.

This logic suggests that an index of sticky prices should be a good indicator of price trends that are expected to persist. The reason is that firms are not going to move prices up or down in response to changing market conditions unless they are pretty sure those conditions are going to last.

So, what has been happening to sticky prices? As the following chart shows, their rate of increase has been slowing since the beginning of the year. If sticky prices really are a valid indicator of future price trends, they are just one more reason for policy makers at the Fed to shift toward stimulus at their next meeting.

3 Responses to “It’s Unanimous: All Indicators Show Inflation is Slowing, Even the Index of Sticky Prices”

Mark StamatakosAugust 25th, 2012 at 2:31 am


They way CPI is calculated today to me is deliberately deceptive. Food costs, energy costs, health care costs, tuition costs, property taxes, and a host of other things we buy or spend for have risen way over the 2% threshold in the last 10 years. I get that some products have become cheaper to own due to advances in technology, but do these really counteract the other price spikes going on?

Do you know anything about the way CPI was reformulated back during Clinton's second term ( I assume you do)? Do you agree with Substitution, weighting, and Hedonics as methods used to track real prices, or are we better off when a basket of food could be measured against a basket of the same food year to year.

I think if the government agencies that deliver data to us actually reported it the way it should be, we'd have:

A higher Federal Budget deficit
Higher unemployment
Much Worse Inflation number
and Substantially lower GDP numbers.

Great site. Hopefully you can respond. Thank you sir.

Ed Dolan EdDolanAugust 25th, 2012 at 1:01 pm

You raise some very important questions regarding the way the CPI is calculated. The CPI is not the be-all and end-all of inflation indicators, but I don't think it is deliberately deceptive. Here are some very brief answers to your questions. I will try to find time to do a longer post expanding on these issues.

1. It is natural for people to distrust government, yet not all parts of government are equally deserving of mistrust. With regard to the BLS, I suggest that you read the views of Keith Hall, who served as BLS commissioner under both Bush and Obama. He says (in private correspondence) "The statistical agencies are comprised of real professionals that do their best given the challenges of collecting data and of tight budgets. It really diminishes the value of the data when people don’t trust it or understand it well." More of his views at

2. Substitution effects (people buy more of goods whose price goes down and less of those whose price goes up) are a problem for all price indexes. Without adjustment, they cause upward bias of indexes like the CPI that use base year weights. To see how big the effect is, go to and compare the normal CPI with the so-called chained CPI; the latter is supposed to make a better correction for substitution effects.

3. Weighting is also a problem. You have to use some kind of weighting; the importance of price of apple juice is obviously not as important as that for gasoline. The best way to get a feel for it is to look at price indexes that use different kinds of weights, e.g., CPI vs. implicit deflator for consumption in the GDP accounts. They track pretty closely, but not exactly.

4. Any price index that did not make quality adjustments would far overstate inflation. Check out this link for an intuitive discussion of the issue:
For a more technical discussion, here is a good article on hedonic methods. They are interesting, but not a deceptive source of large bias. They are used only for a few goods, as the article explains.

5. Much of the popular impression that the CPI way understates inflation is based on the difference between perceived inflation and measured inflation. See this link for a detailed discussion:

Thanks for your questions