Comment by reaperman

Here's a much more detailed analysis of the effects of the executive order: https://chatgpt.com/share/67ee10c6-4690-8006-83d7-8e9b22bccf...

The practical takeaway is that the average household will spend $3,500-4,000 more as a result of the tariffs. Clothes, furniture, toys will be about 30% more expensive, electronics will be about 25% more expensive, tires and jewelry about 15% more expensive, and industrial buyers are going to get hosed when they buy equipment.

There were major carve-outs for the auto industry (yay Michigan) and petrochemical industries (yay Gulf Coast); they’ll still get hosed on equipment but will mostly escape increases in cost of materials. Imported cars/trucks aren’t directly affected either.

------------------------------------

Meta discussion:

I can't do an analysis on the de minimis situation, I don't know of any public datasets that would allow such an analysis, and it's obvious that it will have extremely complex effects (and therefore, any first-order analysis would be very low-quality).

Note on ChatGPT-4.5 "Deep Research": I spot-checked the calculations for HS codes using my own research and the numbers seemed reasonably close to my own. https://atlas.hks.harvard.edu/explore/treemap is an invaluable resource for this kind of analysis. ChatGPT fumbled the bag on HS 30: Pharmaceutical products, by not excluding products listed in Annex II, which overestimated total tariffs by about 6% ($30 billion), but the net effect on households is still in the right ballpark (+/- 20%).

Edit: This isn't one of those simple low-effort posts that say "omg look what ChatGPT said". I'm capable of doing this analysis on my own, and I checked ChatGPT's work for the largest contributing categories of goods. Sometimes we disagreed by 10% or so, but overall the results checked out except for Pharmaceuticals, which I caught and didn't repeat misinformation in my "practical effects" tl;dr. The only way it is significantly wrong is if both myself and ChatGPT missed some large tariff category and assumed it was small - that could raise the real costs above the numbers in the analysis - but I checked all the largest categories that ChatGPT identified as well as all the largest categories from our largest trade partners shown by the Atlas of Economic Complexity, so it's somewhat unlikely that happened.

I purposefully didn't include 2nd and 3rd order effects (e.g. chained CPI) because they are usually relatively small, massively uncertain (my analysis would be worth as much as dog poop on a shoe) and those higher-order effects take too long to manifest. It's not worth predicting costs out 2+ years because the political environment is far too unstable. Just today, the U.S. Senate voted to block all the Canadian tariffs and end the "state-of-emergency" that this executive order is standing on. Front-page diplomacy or private back-room deals could make Trump adjust tariffs up or down on various nations multiple times this year. Who knows what the situation will be 6 months from now, let alone 5 years from now when global supply chains and pricing elasticity might re-normalize. The fact that ChatGPT didn't include these effects is a point in its favor, not a glaring oversight.

I'm not complaining about downvotes but if you are downvoting this, please let me know why so that I can improve. I put a lot of work into this far beyond just typing some crap into ChatGPT and I think these numbers add a lot to the discussion.