Does anyone has experience / can refer to a "general", "data-driven" PARTIAL equilibrium model?

Hello, I am the maintainer of FFSM (, a bio-economic model that deals with forest biological dynamics, forest management and the markets of timber products.
We are currently exploring the possibility to improve, and generalise, the partial equilibrium model behind the market module.

We are looking for a partial equilibrium modelling approach that is independent of the “structure” that we give to the model but is instead “data-driven”: we first observe the market of interest (across a panel of regions/years), and then it is this empirical step that indicates from which other products one particular product depends from, or from which other products its demand competes with.
Ideally, we would then have a single set of “products” with the equation of their supply/demand depending on the prices of all the other products, with a very sparse matrix where few elements would be nonzeros. A non-working implementation of the second step (the partial equilibrium modelling) is in the Markets.jl package.
To be clear, we aren’t looking at a computable general equilibrium problem. We want to remain in the realm of a partial equilibrium approach, we don’t think that the timber market is large enough that macroeconomic feedback should be accounted for.
We are currently investigating the literature for how others have come across this problem of a “general” approach to (computable) partial equilibrium models, that is in no way specific to timber markets, and we would be delighted if some of you in this list could point us to relevant sources or offer some example or suggestion.
Either the idea is valid, and someone must have already implemented it, or for some reasons I don’t currently master it is impracticable, and I would like to understand why :slight_smile: