When we talk about the effect of some policy change—rate cuts, tariffs, a tax increase, etc.—the simplest way to do so is to take a very mechanical approach. Rates get cut, so money is cheaper, so projects that weren't quite high-return enough to merit investment can get it (or families that would only move into a house if their monthly mortgage payment were below some fixed amount will do so). And some people do behave this way, especially if they're cash-constrained.

On the other hand, capital allocators are not exactly shocked that the Fed periodically meets and adjusts rates, especially if these adjustments are telegraphed in advance  (see: forward guidance) and respond to specific macro datapoints. If there's a time when the market is expecting a cut, and that cut doesn't happen, they tend to shift their assumptions to expecting a cut at the next meeting. And if there's an investment that makes sense when rates are a quarter-point lower, they'll make it once they're convinced that that's where rates are going, not when it actually happens (and their lending counterparties, who are aware of the same data, will also price things accordingly). For example, a cut late last year sent long-term treasury bond yields up, even though short-term rates went down; the market's read was that this rate cut was likely to spark inflation, and that the real value of a long-term treasury bond was lower in that scenario.

Rational expectations matter a lot for inflation specifically. If consumers think prices are rising, they'll tend to buy more durables, buy in bulk, borrow to buy houses, etc. And, of course, all of that activity increases inflation—more durables have to be built, shipped, and sold; more bulk products have to be manufactured; housing supply has to expand or its price will go up. These consumers will also be leery of longer-term fixed-income, and may shift money out of bonds and into assets that handle inflation better.¹

This also shows up in the response to tariffs. Prices on a fairly random assortment of goods have been raised, and now their pricing is less certain. But:

1. Companies knew tariffs were coming when the guy who calls himself Tariff Man won the election. Conveniently for these companies, there tends to be a shipping rush and warehouse space crunch in Q4, because of holidays, which means there's slack in both systems in Q1. So they had time to stock up.

2. Once tariffs hit, the numbers were so absurdly high compared to expectations that many companies bet that they'd be lowered, or there would be broad exemptions, or they'd be struck down by courts, or something would happen so Vietnamese exports wouldn't have their prices hiked by 46% or something.

Companies could have a simplistic policy of marking everything up based on cost, and suddenly raising the prices when they burned through pre-tariff inventory. They could also have a blanket policy of marking up based on marginal cost, and pricing everything they sold based on what it would cost to order the next replacement. But they tended to be a little more sophisticated, and to smooth out the spiky impact of policy by predicting its arrival and, once that happened, predicting its imminent departure.

But this is where rational expectations get too cute for their own good. If you always behave as if you're living in a reasonable world, then the unreasonability of the real world imposes constant drag on your returns. So you typically have to respond at least a bit to policy changes, even if you don't think they're likely to stick around.

But what makes this really challenging is that the policy response is partly a function of the economic response. Tariffs would be rolled back faster, and probably be a political taboo for a generation, if they'd fully gone into effect and unemployment had promptly spiked to 8% while equities lost a third of their value. Equities did drop, and the job market is wobbly, but stocks promptly rallied on what amounts to a real expectations tariff thesis (helped along by a Real High Expectations for AI thesis), and instead, we get things like GM beating earnings estimates because, thanks to policy tweaks and their own rerouting of supplies, the impact of tariffs is less bad than expected rather than good.²   GM shareholders will be happy to take that, and shares popped on the news, but every tariff-mitigation rally implies that the net present value of future tariff expenses is now higher since there's less pressure to reverse course.

As this newsletter has pointed out before, the real impact of policy shifts is not who is immediately made better- or worse-off based on decisions they made before; it's what decisions they'll make differently in response. If corporate tax rates decline, there's a temporary windfall to companies, but if companies respond by expanding, and they compete with one another, then this effect washes out and the real impact is a shift from consumption to investment. If rates drop, money only gets cheaper to borrow to the extent that lenders think rates should have dropped, and those lenders should already be pricing things based on where they think rates ought to be—if you think rates are too high, and you can lend at these too-high rates, it's your fiduciary duty to go hog-wild first and not wait for the Fed to ratify your observations. But all of this means that actually determining the impact of policies is only possible when they're surprises. And when it comes to macroeconomic policy, it's a lot easier to think up bad surprises than good ones.

We've written about rational expectations implicitly many times in The Diff, including thoughts on:

• How tariffs have historically flowed through to FX rates, blunting their impact ($).

• The long-run effects of tariffs ($), including a modest proposal where countries that run a surplus can offset the impact of tariffs domestically.

• Why the US electorate likes deflation ($).

• How the Covid inflation mimicked postware inflationary episodes ($).

Join the discussion!

1  Equities seem like they'd be able to adjust. After all, aggregate inflation is the average price of things people buy, and so it means that revenue is higher for the sellers. But there are a few problems with this. First, if inflation isn't uniform, it's possible that some companies will have a spike in costs that they can't fully pass on to consumers, and others will have comparatively smaller incremental costs and do smaller price increases. Second, these price increases are expensive, and consumers tend to partly attribute inflation to the greed of whoever last marked something up. And third, there are frictional costs to raising prices, though in some cases, like fast food, selling products through apps and aggressively price-discriminating with loyalty points means that they can more seamlessly adjust prices and that customers are less likely to notice ($, Diff). If the frequency of temporary discounts on chicken McNuggets drops 20%, it's a lot less noticeable than if the price rises.

2  It’s interesting that GM was likely able to reroute this supply partially as a result of their access to AI via Palantir. Sometimes the themes driving the market are so big that the can’t help but overlap.