Federal Data and Its Discontents Loom Large as Shutdown Wears On

As America’s partial government shutdown drags on, it’s depriving economists and analysts of their intellectual lifeblood: federal data.

To understand why government data is so crucial, it’s worth looking back to a time before comprehensive federal statistics existed — and fortunately Hugh Rockoff at Rutgers University has a National Bureau of Economic Research working paper out on just that point.

The U.S. has collected census data since 1790, but in Rockoff’s telling, most government indicators emerged as an answer to a Great Depression-era problem: how do you know how income is distributed and how Americans are faring without a central information repository?

Commerce Department statistics on gross domestic product, for instance, have their roots in the 1929-1932 crisis, and “the political point, clearly, was to justify sweeping governmental economic initiatives,” Rockoff writes.

The first national income report — submitted by Simon Kuznets in 1934 — served as a call-to-action that helped the government understand that it needed to respond aggressively to the downturn.

While expectations that federal statistics can accurately predict turns in the business cycle often fall short, Rockoff points out that today,“most private business plans depend in some measure on ideas about where the economy is headed.”

On that note, the fact that the Commerce Department and Census Bureau lack approved spending — holding up reports from retail sales to inflation — is sending Main Street companies and Wall Street firms into a scramble for some way to assess the economy.

That’s led analysts to tout their own private indexes. Morgan Stanley is holding up its output activity index, which shows signs of a moderating economy, as a proxy for official GDP data. Organisation for Economic Cooperation and Development figures fill other gaps, analysts say.

Copyright 2021 Bloomberg. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.