Data revisions are a normal, necessary part of trying to measure a $30 trillion economy in real time. They can also be jarring, especially for people who don't obsessively scrutinize every jobs report. So I dug into the how and why of revisions. A few big takeaways:

1. No, there's no evidence of political bias. Period. This lovely chart from @collinskeith makes that clear: There were downward revisions under Biden and Trump.

See also this earlier thread debunking the claim that BLS revised the jobs numbers down after Trump won. (In fact, they revised them down *during the DNC*, hardly an opportune political moment for Biden/Harris.)

2. Revisions are essential. All economic statistics represent a tradeoff between timeliness and accuracy. If you don't revise, then you just have to pick one -- timely or accurate. Revisions, as @JedKolko explained, are how you can get both.

4. However, the pattern of consistent downward revisions is unusual. (Again, though, it predates Trump.) Ordinarily, downward revisions are a bad sign for the economy -- the initial estimates tend to lag at turning points.

But in this case, there may be other explanations. We've seen big shifts in immigration, entrepreneurship, seasonal patterns, etc., which could be affecting the BLS's models.

And the revisions on Friday were concentrated in local government, as this chart from @christinezhang shows. That could reflect seasonal adjustment or other issues.

5. BLS, like other statistical agencies around the world, *does* have big issues with declining survey response rates. That hasn't obviously affected the quality of its data yet, but if the trend continues it will.

Experts both inside and outside BLS have acknowledged it is going to need to rely less on survey-based data sources and incorporate more administrative an private data. They've taken steps in that direction already.

But changing methodologies and incorporating new sources is resource-intensive (even if it might cost less in the long run). The statistical agencies have seen their budgets shrink for years, and the problem has grown worse under the Trump administration.

As Nancy Potok, the former chief statistician of the U.S. in the first Trump admin, put it: “The solution to that is not to fire the head of BLS but to make the immediate investments in the federal statistical system."

Much more in the story. Free link: