Has there been research on algorithmic/dynamic tariff optimization?

I keep coming back to this idea and I’m not sure if it’s naïve or just underexplored: what if tariff policy worked more like monetary policy? Not discretionary politics every few years, but something closer to a rule-based system that reacts to data.
Trade policy is always pulling in opposite directions. You need foreign competition so domestic firms don’t turn lazy and rent-seeking. But too much competition collapses margins, kills long-term investment, and eventually wipes out capacity and R&D. At the same time you’re juggling consumer prices, employment, strategic independence in certain industries, and the hope of pushing the economy up the value chain. Right now we mostly handle this with blunt, static tariffs and political bargaining, which feels… crude.

What I’m wondering is whether anyone has seriously studied a dynamic system. Imagine tracking a small set of real indicators in near real time: sector-level profit margins, domestic R&D spending, capex, prices, employment, estimates of foreign subsidies, maybe even strategic capacity ratios. Then tariffs adjust automatically within preset bounds. If the system detects heavy foreign subsidies and falling domestic investment, it applies a countervailing tariff—say 60–70% of the estimated subsidy—enough to preserve competition without letting domestic firms coast. If domestic investment picks up and the price gap narrows, the tariff phases down on its own.
This feels fundamentally different from Soviet planning to me. You’re not micromanaging production or prices. You’re only touching one variable at the border, while letting internal markets stay free. It’s closer to how central banks move interest rates without trying to control every transaction in the economy. Prices, profits, and investment still do the signaling; the policy just nudges the environment they operate in.
That said, I can see all the red flags. This runs straight into the socialist calculation problem, just in a narrower form. Who sets the objective function? How do you weight prices versus innovation versus employment? How do you stop firms from gaming the metrics or capturing the process politically? And even if the information requirements are far smaller than full central planning, are they still unrealistically demanding?
One thing I’m especially stuck on is how you’d distinguish natural comparative advantage from artificial ones. If a country is genuinely better at producing something because of geography, climate, or decades of accumulated know-how, tariffs shouldn’t interfere with that. The system should only counter distortions coming from subsidies, dumping, or other policy-induced advantages. Related to that: how do you tell when competition is healthy and innovation-driving versus when it’s so intense that it just destroys investment incentives altogether?

I know about computable general equilibrium models, but they seem way too heavy and slow for anything resembling real-time adjustment. I’m curious whether there are papers, computational models, or historical experiments that look at this narrower idea: dynamic, feedback-driven trade policy that tries to stabilize incentives rather than pick winners.
If this is a bad idea, I’d like to know why in a non-handwavy way. And if it’s been explored before, I’d love pointers to the literature or cases I’m missing.