Harnessing the Power of Big Data with CARTS
August 9, 2021
By Scott A. Brave, Senior Economist, Federal Reserve Bank of Chicago
The term “big data” means different things to different people. To me, big data refers to the rapidly growing and constantly changing frontier of information collection and data analytics, accumulated mostly by the private sector.
As the pandemic unfolded, the speed with which the economy responded to the spread of the virus and the public policy actions that followed laid bare the need for accurate and timely measures of economic activity for governmental policymakers. This was particularly true for measures of consumer spending—which helped to track the effects of the Covid-19 crisis on business (and government) revenues.
My colleagues at the Chicago Fed and I have spent the better part of the past year learning about and coming up with novel ways to make use of big data for this purpose. Recently, we began publicly releasing one of the fruits of our efforts: an index called CARTS—which is short for Chicago Fed Advance Retail Trade Summary. What is this index and how can it be used to study changes in consumer behavior in real time? Here are the answers.
What is CARTS?
CARTS is a weekly index of national retail spending that is designed to track the U.S. Census Bureau's Monthly Retail Trade Survey (MRTS). The MRTS is a much-used snapshot of consumer spending on goods and select services and is used by multiple government agencies for purposes such as calculating gross domestic product and analyzing current economic activity and trends in consumer purchases.
The problem with the MRTS, however, is that it only comes up with data on a monthly basis, while state and local governments are generally better served with information that can be tracked on a more frequent basis. Big data can reveal trends that are helpful in making decisions related to revenues and expenditures. CARTS helps to fill in the information gap left when relying on the MRTS alone by using high-frequency data from five private companies (Consumer Edge, Facteus, Morning Consult, SafeGraph, and Womply), as well as two federal agencies (the U.S. Energy Information Administration and the U.S. Census Bureau). Together, they provide an early and reliable snapshot of weekly retail spending based on credit & debit card transactions, retail foot traffic, gasoline sales, and consumer sentiment.
Reconciling private big data sources with public official statistics can be very difficult, but CARTS was designed with this challenge in mind. By using multiple sources of information, CARTS reveals substantial week-to-week variation that often cannot be detected in the monthly MRTS. Unlike its underlying data, CARTS is also benchmarked to the MRTS. Because of this, the historical values of CARTS, when averaged over the month, are identical to the MRTS measure of retail & food services sales excluding motor vehicles & parts.
What can I learn from CARTS?
When the MRTS measure of retail & food services sales excluding motor vehicles & parts is not yet available for the latest month, CARTS estimates it. Since March 2020, the CARTS projection of this data series has been roughly 50% to 75% more accurate than either consensus forecasts available at the time or other common forecasting models that rely solely on the monthly MRTS data.
The weekly frequency of observation of CARTS allows for the study of the high-frequency drivers of consumer spending at a granular level. For example, CARTS reveals that both private big data sources and official public data sources play important roles in explaining high-frequency fluctuations in national retail spending. Indeed, the use of multiple types and frequencies of data on consumer spending is an advantage that CARTS has over other existing alternatives.
CARTS tracks national retail spending with less volatility than several alternatives that rely solely on credit & debit card data. At the same time, it also reveals facets of national retail spending that cannot be seen in the MRTS. For example, CARTS reveals a stockpiling effect in March 2020 just prior to the widespread implementation of lockdown measures across the country, which is not noticeable in the averaged monthly spending data from the MRTS.
Additional examples of what CARTS has the capacity to do include tracking the real-time effects on consumer spending from the rise and fall of Covid-19 cases in the U.S. and from the disbursements of economic impact payments. Furthermore, while CARTS is a national index, its methodology is flexible enough that in the future it could also be applied to state-level data as well.
Where can I find CARTS?
The Chicago Fed releases updates to CARTS that are available at https://chicagofed.org/carts at 10:00 a.m. ET on scheduled days. The next release date falls at the end of this week: Friday, August 13, 2021. Note that while CARTS is a weekly index, it is released only twice per month: The first release of the month covers the entirety of the previous month (final release), and the second release covers the first half of the current month (preliminary release).