Updates & Erasures:
Another week, another funky datacenter funding vehicle, aka an SPV. The latest example of "hide the AI datacenter capex", which I've been on about repeatedly, comes from xAI. It is working with Valor, the WSJ says, trying to help it raise money to buy more chips and such for Grok-type things. And that, it seems requires an SPV.
Why? Well, the tricky bit requiring the SPV, other than it's fun to hide debt off-balance sheet is that xAI apparently has debt covenants prevented it from raising more than an additional $5b. But with $12b (what it wants to raise) being materially larger than $5b (what it is allowed to raise), there is a problem. Bankers, being nothing if not creative, enter the idea of using a legal artifice to attach the debt to something that is not xAI, but is, for all practical purposes.
This will all work out well, boosting GDP and such, of course, until it doesn't.
Rough Notes:
Tariffs Are Just Stupid Enough to (Almost) Work
Tariffs are dumb. They distort trade, favor inefficient local producers, cause trading partners to retaliate, and make people worse off than a world without them. On these points, economists almost universally agree¹.
¹ This is weak. Saying that "most economists agree on XYZ" can be like saying that it is common sense, like backing out of your driveway quickly is a bad idea. Totally, yes, 100%. Didn't need you to tell me though, etc.
But tariffs are not useless. They may even be sort of, almost, kinda, a ... good idea in these very weird U.S. circumstances.
Hear me out, because three things are going on, so it can get messy:
- The U.S. is, as the line goes, an insurance company with an army, which has straitjacketed its budget, which I've written about previously.
- The U.S. hates taxes, and most voters are innumerate, so it finds silly ways to hide them.
- Tariffs are a kind of horrible, second-best solution to the above problems.
The first two points are mostly self-explanatory. Entitlements plus defence are now around 70% of the U.S. budget—see also, insurance company with an army—leaving little room to do much other than cut, unless you find new revenue. But new revenue is hard, because Americans hate income taxes, and have long resisted carbon taxes or a value-added tax (VAT). They aren't coping well with what I've called life under 2%.
Enter tariffs. They raise money because consumers buy things. We can argue about whether the producing companies pay the tariff (they mostly don't), or whether consumers pay it via higher prices (they mostly do), but the effect is the same: consumers buying things increases government revenue. That is tariff income.
So far, so ... suboptimal. Because tariffs aren't a good tool for this. I will come to why they aren't very good in a few paragraphs, but they distort, create weird incentives, invite retaliation, etc.
A much better tool is a value-added tax (VAT), a broad tax applied to consumer purchases of goods and services. Most countries have one, including all of the OECD except for the U.S.
It is generally agreed that VATs are a good idea, that they can be less distorting than income taxes. And, most importantly, if you're a government, they produce gobs of income for countries that have them. How much income? The average nation's VAT income is around 6% of GDP.
So, why doesn't the U.S. have a VAT of its own? After all, the country has what are often obfuscated as significant long-term fiscal challenges. These mostly revolve around trying to run a costly modern social democracy on a low-tax system. This mathematically intractable "challenge" is made worse by a healthcare system unrivaled for all the looting intermediaries demanding to be seen instead as paragons of competition and capitalism.
There are various reasons for having no U.S. VAT, but the most important is in the name: it is a tax. And Americans hate taxes. Just ask them. The U.S. government cheerily indulges them in their hatred of taxes by cutting the taxes they can see, like income taxes, and hiding the ones they can't, like the pre-tax corporate deductibility of healthcare premiums (costing $300b and 1.5% of GDP). This has costly & malign effects, like a 6+% structural budgetary deficit and the most screwed-up and expensive healthcare system in the world.
How much money is the U.S. foregoing by not having a VAT? It is large. Here is the rough math:
The Assumptions:
- U.S. GDP (2024): ~$28 trillion
- VAT base (~50% of GDP): ~$14 trillion
- OECD average VAT rate: 20%
The Math:
- Possible U.S. VAT revenue: $14T×20%=$2.8T
- Current VAT revenue: $0
- Foregone Revenue: $0 - $2.8 trillion = $2.8 trillion
The U.S. is foregoing approximately $2.8 trillion annually in potential VAT revenue at an OECD-average rate. Even at half that rate—because, America!—a U.S. VAT might produce, all else equal², around $1.4 trillion a year.
² Among the more pedantic people on earth are tax policy types. They will explain, at length, that all else isn't equal, that VATs have leakage, avoidance, static vs dynamic effects, and more, all preventing them from raising as much money as one might hope. All agree, however, that having a VAT raises far more money than not having one. Like, all of it.
To put that in a kind of context, the current U.S. budget deficit is around $1.8-trillion a year. A VAT set at even half of OECD average levels would nearly zero out the U.S. deficit. (And, of course, reforming U.S. healthcare by eliminating premium pre-tax deductibility, instituting universal Medicare Lite, and requiring catastrophe insurance would flip the U.S. to surpluses, but I digress.)
Let's now turn to tariffs. Like a VAT, they are broad consumption taxes, just not applied defensibly. They are applied only to imports, not to everything bought and sold in the country. This makes no sense, unless you think tariffs aren't taxes (they are), and you think tariffed companies pay them (they don't). So, Americans.
But tariffs are a species of VAT, albeit a poorly designed one. A universal tariff on imported goods—say, at 15%—would raise VAT-lite revenues. Based on recent data, U.S. annual imports are around $4 trillion. Applying a uniform 15% tariff to manufactured goods, which is 80-ish% of that. might yield roughly $300-$400 billion annually. While this is a fraction of the revenue of an actual VAT, it is real money. The choice then is not between a perfect VAT and an imperfect tariff, but between an imperfect tariff and continued reliance on deficit financing or distortionary taxes on labor and capital income.
Whoa, whoa, whoa, you might rightly protest. This is just a bad solution. Sure, but it is, in practical terms, a "second-best solution", even if it is also perhaps the second-worst.
We should want more second-best solutions, economics tells us, if the alternative is doing nothing. There is a framework³, with which I won't bore you, that says it's okay to do something less than perfect, if by doing so you counteract some of the problems preventing you from doing the best thing.
³ Economists Richard Lipsey and Kelvin Lancaster showed in 1956 that if you can't do the optimal policy then the next-best policy might not be so bad. If you can't fix the distortion preventing you from doing the best thing, then adding in some more distortions may kinda offset the first one, and lead to a more efficient outcome. This can seem like adding cayenne pepper to pasta sauce to offset too much salt, admittedly.
In this case, American politics prevents an actual VAT from happening, so perhaps tariffs aren't so bad, if the alternative real distortion is structural deficits. To that way of thinking, distorting trade via a uniform tariff (a second distortion) may increase overall welfare relative to the status quo (deficits), despite being shitty trade policy.
And, if we want to spitball here, tariffs could even lay the groundwork politically and psychologically for a future transition to an actual big-boy VAT. Citizens and businesses might recognize that consumption taxation you can see is better than consumption taxation that you can't. A future administration could leverage dissatisfaction with tariffs to propose replacing them with a more economically efficient and lower-rate VAT. Politically, the VAT would then become not a "new" tax but rather a tax cut (in rate terms only) eliminating import tariffs.
The debate over tariffs versus VATs is about the current structural problem in U.S. budget, a refusal to recognize life under 2%. Economically ideal policies frequently fail politically, leaving policymakers with second-best solutions. Tariffs, undeniably flawed and distortionary, are a usefully ugly compromise. They generate meaningful revenue, shift some production domestically, and potentially serve as a stepping-stone toward a VAT.
Hey, a boy can dream, can't he?
Sketches:
Junk food brands are under pressure as consumers pull back on crap … AI labs’ all-or-nothing race leaves no time to fuss about safety … Long-term shifts in heart disease mortality show regional divergence … Private equity firms flipping assets to themselves in record numbers … Norovirus vaccine prospects emerge amid testing breakthroughs … Who killed the miners of Buffelsfontein in South Africa’s gold fields? … Geoffrey Hinton on AI vs. human intelligence and future risks …Western missile demand drives U.S. weapons makers’ growth … Relationship recession goes global amid shifting social norms … What if AI made economic growth explode globally? … Do superstars perform after job moves or fade? … Jamie Dimon warns on private credit while JPMorgan expands into it … Health futures on a damaged planet intertwine with climate risk … IKEA’s prints transformed home aesthetics globally … Predicting penalty kicks with human action recognition models … Early European technology revisited through archeological biology … British state grapples with catastrophic Afghan data leak … Industrial pivot to data centres amid AI infrastructure buildout … AI-driven electricity demand may raise U.S. power costs … The sun’s new significance in climate and solar science ... Ghosting is out, speed-dumping is in
