JPMorgan Economists See Promise and Pitfalls in Alternative Data
(Bloomberg) -- Alternative economic data collected from new technologies such as Google Trends can offer visibility into activity that traditional measures miss, and with a shorter lag, especially for areas where information is scarce or unreliable, according to JPMorgan Chase & Co.
But there are also big challenges such as short histories, collection systems that are prone to change and inconsistent samplings of the population, according to a report Wednesday from economists Jesse Edgerton and Dan Weitzenfeld. For example, they find some Google Trends Internet-search data correlate strongly with U.S. retail sales, but more so for longer-term data than seasonally-adjusted monthly measures Wall Street forecasters often watch.
“Despite these challenges, we still think there can be substantial benefits from using alternative data,” the analysts wrote. “The benefits are likely to be clearest in emerging or other economies that have fewer available data than the U.S., and for answering more granular questions about specific firms, industries, demographic groups or geographic regions.”
The study is the latest in the long debate among economists and investors about the value of new sources of near-instantaneous data about consumers and economic activity. While real-time data from mobile phone locations, online job ads, real estate listings and satellite images can look tantalizing to investors looking for an edge, government reports and business surveys remain the bedrock for following the economy -- even if they’re slower.
The alternative data could prove useful as the ongoing partial government shutdown delays the release of official data. Fed Chairman Jerome Powell said last week that a prolonged shutdown could have an impact on the U.S. economy, while the absence of all the government data would would obscure the economic outlook for policy makers.
Edgerton and Weitzenfeld cited work by stock analysts at the bank who use data compiled from satellite imagery tracking car counts in parking lots to inform ratings on retailers, and the JPMorgan Chase Institute’s use of the company’s own credit-card and banking data to better understand spending in individual cities and by small businesses.
More granular data may aid so-called “bottom-up” economic forecasts, but there are only modest prospects for alternative data to improve U.S. macroeconomic predictions, in part because they’re volatile, they said. Daily high-quality alternative data have benefits because they’re more timely and offer a “second opinion on official data releases” such as retail sales, which are subject to errors and biases, they said.
Such data could help smooth noise in official data, and “may prove most useful in detecting true turning points days or weeks more quickly than the official data,” they said. “We would be happy to examine data from any alternative data providers who find our conclusions too pessimistic.”
Federal Reserve officials have used new tools to track consumer spending, such as credit-card swipe information from global payment processor First Data Corp. And the U.S. Bureau of Economic Analysis has been experimenting with hundreds of different prediction models that use public and private data, such as Google search queries, to help improve their estimates for service-sector consumption that are included in gross domestic product.
Goldman Sachs Group Inc. also sees a more nuanced outlook for alternative indicators. Despite more inclusion in official government indicators, statistical agencies still face significant constraints ranging from budget pressures to the legal authority for access, economist Spencer Hill wrote in a report last May.
“In terms of the mainstream approach to tracking the economy, it sometimes feels like nothing has changed,” Hill wrote. “To paraphrase Charles de Gaulle: Big Data is the future of macro... and always will be.”
©2019 Bloomberg L.P.