Tomorrow morning, @BEA_News will release its prelim. estimate of second-quarter gross domestic product. Not much drama on the headline: It will show a historic decline in economic output.

But the way the numbers are presented may sow confusion. (thread) https://www.nytimes.com/2020/07/29/business/economy/us-gdp-report.html

Forecasters expect the report to show that G.D.P. fell at an annual rate of about 35% in Q2.
If you don't follow econ data closely (or maybe even if you do), you might have read that as saying output fell by more than a third in just three months.
BUT THAT WOULD BE WRONG

In the U.S., G.D.P. isn't reported as a simple change from one quarter to the next. It's reported as an annual rate (technically a seasonally adjusted annual rate). Think of it as "if this rate of change held steady, this is how much G.D.P. would grow/shrink over a full year."

If tomorrow's report shows a -35% annual growth rate, that means economic output was *10.2%* smaller in the second quarter than in the first.
(It's not as simple as dividing the annual rate by four, because growth rates compound.)

Looking at things on an annualized basis can be useful, because it makes it easy to compare data collected over different time periods. You've probably done a version of this calculation yourself! "OK, if I cut out my $4 daily latte, that would save me $1,000 a year..."

But when annual rates get applied to short-term or one-off changes, they can be misleading. You'd never think of a $500 holiday bonus as "$6,000 on an annualized basis."

Right now, the economy is experiencing *a lot* of short-term swings. Businesses shut down in April. Many reopened in May/June. Now activity may be slowing again again. Those developments are real and important -- but it doesn't make much sense to annualize them.

For that reason, in our coverage tomorrow, the Times plans to emphasize the simple, nonannualized change from the first quarter to the second. We'll still give the annualized number for those used to seeing it that way, plus other numbers as relevant (year-over-year change, etc).

This approach will make the Q2 change look milder than if we used the annualized figure. We plan to do the same thing next Q, when it will (presumably) make the rebound look smaller. But in both cases, we believe it will more clearly communicate what is happening in the economy.

Note that in charts, there are other interesting options: showing *levels* as opposed to change, or cumulative change from a given starting point (as @ellawinthrop has been doing with our jobs charts lately). We're still exploring our options. https://www.nytimes.com/2020/07/02/business/economy/jobs-unemployment-coronavirus.html

We're making this decision because we believe it's what will best serve our specific set of readers. Other news organizations, with other audiences, may reach other decisions. But I hope that all of us who communicate economic data to the public are thinking about these issues.