Computing the Cost of Intelligence: The Consumer Price Index and Information Products

Every month the U.S. Bureau of Labor Statistics (BLS) announces its latest estimate of the cost of all consumer goods and services. These initial estimates are then revised as additional data come in. One of the key drivers of the American economy over the last few years has been the declining cost of information in relation not only to other goods and services but occasionally also in nominal prices. This blog reviews the current procedures for measuring information costs, hopefully giving readers more context about the long-term trends.

We examine seven discrete categories of goods and services that have the primary purpose of collecting and processing information. Four of these goods are members of the broader category of Information technology, hardware and services. Two lie within the category of Telephone services. The final one lies within the category of Video and audio.

Approximately half of the index for telephone hardware, calculators, and other consumer information items consists of cellphones (about 95 percent of which are smartphones) with home-based phones, phone accessories and smartwatches making up the rest. The Consumer Price Index (CPI) records the actual price of a smartphone, with promotions deducted from the advertised price. The BLS also separates out the price of hardware from accompanying services such as Internet access and television subscriptions. These services are collected in a separate category of spending.

One of the most important information products is smart phones. These goods are the only item in their broader category that are quality adjusted due to the rapid rate of technological advancements and improved quality for consumers. For the vast majority of products, BLS simply records the price without any adjustment for quality. However, for smartphones BLS economists have developed a hedonic regression model that allows them to identify how much of a price rise is due to specific improvements in quality. Assuming some of a price increase was due to product improvements such as better screen resolution, BLS would add the value of those improvements to the previous price of the smartphone and compare it to the new price. If both prices were the same, consumers would be paying more but also getting more, as opposed to paying more for the same benefit due to general inflation. Wireless telephone services and television services undergo similar adjustments.

CPI also uses the concept of directed substitution for both smartphones and computers because of the rapid quality improvements in each good. Basically, twice a year BLS assumes that consumers are purchasing the newer version of a product even if the older one is still available. The resulting data recognizes that manufacturers are improving the cost and/or quality of the product even if consumers are not purchasing it yet.

Another key fact is that information services make up a growing share of consumers’ total spending. This multiplies the impact of cost restraint. A 50 percent price reduction in housing, which currently accounts for more than 44 percent of spending, has a much larger effect on living standards than a 50 percent cut in furniture and bedding, which makes up less than 1 percent. Table 1 shows the relative weights for selected categories of information spending for 2017-18 compared to 2024. Some items have increased their share of the consumer’s budget while others have contracted.

Table 1: Relative Importance of Components in the CPI, 2017-18 and 2024

Spending Category	2017-18	2024
Computers, peripherals, and smart home assistants	0.285	0.293
Computer software and accessories	0.016	0.027
Internet services and electronic information providers	0.856	0.923
Telephone hardware, calculators, and other consumer information services	0.066	0.461
Wireless telephone services	1.824	1.340
Residential telephone services	0.405	0.126
Cable, satellite and live streaming television service*	1.145	0.606

*Was Cable and satellite television service in 2017-18

It is important to note that the CPI only collects data on consumer items. Price and quality improvements on products purchased by businesses are not recorded. This excludes a major source of innovation and productivity in the economy. However, many of these changes are eventually captured as competition eventually forces companies to pass most of the benefit on to consumers. Also excluded from this table are subscriptions to on-line magazines, games, and music and video downloads, which are treated just as if the consumer purchased them offline. Non-business subscriber fees for residential television are included, but pre-recorded video-on-demand subscription streaming services are not.

BLS faces a constant problem separating out changes in general inflation from changes in the quality and capacity of specific goods and services. Products that seldom change from one period to another present little problem. However, the true prices of products that change rapidly are more difficult to measure. A current example is the rise in pay-per-view streaming which charges viewers fees for a greater choice of programming content. If a replacement product is better than its predecessors and BLS can measure the value of the difference in quality, it will raise the price from the previous month to reflect that there has been no change in the real price. In cases like this, nominal price increases may simply reflect the fact that consumers are getting more for their money. In this case the market is doing exactly what it should.

The relationship between prices and growth is complex. Some experts argue that much of the U.S. productivity boom in the 1990s and 2000s was due to statistical agencies using hedonic measures to overestimate productivity gains. However, others point to the large unincluded value of free services such as ChatGPT, Facebook, Gmail, Google Maps, and YouTube. Although these services deliver tremendous value to consumers, they are not counted in GDP. This debate will not end soon. In the meantime, the important point is that technology does increase our living standards even if the inherent level of data uncertainty makes this hard to see.