GDPNow First Quarter GDP Estimate Sinks To 2.0 Percent. Still Too High?

Yes, based on trade data and inflation analysis. Explanation below.

The GDPNow Estimate is 2.0 percent as of March 23, 2026.

Understanding GDPNow

GDPNow is a “Nowcast Estimate” not a prediction. It’s a running estimate of what the BEA would report IF the BEA were to produce a GDP report at the current time.

Initial estimates often look and often are ridiculous. That’s not a bug it’s a design feature.

As economic reports come in, the nowcasts tend to stabilize, at least somewhat.

GDPNow March 2023 Numbers

GDP Nowcast: 2.0 percent
GDP Real Final Sales: 1.5 percent
GDP Real Final Private Domestic Sales: 1.8 percent

Real Final Sales vs GDP

The difference between GDP and Real Final Sales (RFS) is Change in Private Inventories that nets to zero over time.

RFS is the actual bottom line number. For that reason, the BEA would be wise to throw the headline number in the garbage.

The Fed focuses on Real Final Private Domestic Sales. It’s currently between RFS and the topline estimate, but it could be the highest or lowest. It just happens to be in the middle now.

What Happened on March 12?

On March 12, the January 2026 full trade report showed a sharply narrower trade deficit ($54.5B vs. ~$66B expected by consensus). Exports surged, imports dipped.

The GDPNow contribution from net exports improved from –0.50 pp to –0.32 pp, and the headline nowcast jumped from 2.2% to 2.7%.

That lift wasn’t because January trade was “strong” in absolute terms — it was because the data came in much better than GDPNow had been projecting.

The model had been carrying a more pessimistic view of net exports heading into that release.

January Trade Data

My view is that January trade data, especially imports, were weak because importers held back waiting for the Supreme Court tariff decision.

On February 20, the Supreme Court ruled against Trump. Then the International Trade Court mandated refunds.

I now expect a surge in imports reversing the January numbers.

Given typical 20–36 day ocean transit + clearance, the real “mad rush” import catch up lands mostly in March data.

This Friday, we see advance trade numbers for February. Nothing will surprise me here. But GDPNow will not use that number

On April 1, we have the full international trade data that GDPNow will look at. That data is for February. So take any GDPNow change on April 1 with a big dose of salt.

I expect the full brunt of imports to hit in March with perhaps a bit of spillover into April. Thus, I expect GDPNow to be skewed on the high side until the full March trade data release.

GDPNow Estimated Contributions to 2026 Q1 GDP

As of March 23, net exports are imports at -1.43 PP + exports at 1.09 PP = -0.34 PP.

GDPNow Contribution for Net Exports

Key Point on Analyzing GDPNow Reports

It’s not the data that matters, but rather what the GDPNow model expected that really matters.

GDPNow did not react to the data on the 12th per se. It reacted to the data vs its expectation of the data. Very few people understand this key point.

This is why people don’t understand how the model often reacts opposite the data reports.

It’s “possible” the model now expects worsening trade. However, the history for net export contribution suggests the model does not expects trade data to worsen.

I do, for reasons stated. Thus I expect the model’s expectation is too high, and that is what really matters.

CPI and PCE Inflation

Higher inflation will reduce real GDP.

Due to the war in Iran, there is going to be a surge in reported inflation.

That’s my second reason to expect a weaker forecast than the current GDPNow estimate.

I doubt the model expects the upcoming inflation reports.

Finally, I expect jobs to continue to weaken, and consumer spending along for the ride. That’s a third reason to suspect the Nowcast is too high.

If so, the combination of trade data, inflation, and weakening spending could easily sink the GDPNow estimate towards zero.

Q: GDPNow Still Too High?
A: Yes, for two or three reasons

Models vs Humans

The problem with models is they can’t think. They are only as good as the human (or AI) making the model.
The problem with humans is they overthink, think about the wrong things, and have expectation biases.

GDPNow is just a tool, but a very good one provided you understand the above points. I appreciate the GDPNow model.