Notebooks

Economic Networks

Last update: 06 May 2026 11:20
First version: 12 April 2025

Production/consumption networks: What one economic entity produces, another consumes; this defines a (directed) graph between entities. At the level of industries, this is input-output analysis. At the level of plants or establishments, there isn't (I think) a settled name for the topic; nor is there at the level of firms. Interestingly, while the firm-level network and the industry-level network are both aggregations of the plant-level network, they are not necessarily nested, because there are firms which are active across multiple industries...

Other sorts of economic networks: buyers and sellers who coordinate through exchanges or brokers (a bipartite graph, or tripartite if buyers and sellers are distinct); choice of technology (keyboard layout, operating system, language...) where utility depends on the size of the network of other adopters (at least in part); credit provision in finance; corporate boards (historically, the first topic of network analysis in the social sciences, back in the late 1800s).

The borders between all these kinds of thing and social networks is fuzzy, so have a link.

Queries: To what extent can macroeconomic fluctuations (i.e., the business cycle) be understood as the result of the network structure of the economy? Can one explain Slutsky-style moving average dynamics as the result of shocks propagating through the network? (I'm sure people have actually worked out stuff like this, or worked out why it's not right, rather than merely having idle thoughts, I just haven't read it yet.) --- Immediately after first posting this, Rachel Childers provided some pointers to the literature, now incorporated below.

Network position and potential networks: It is easy, and common, to look at an actually-existing network and try to determine which nodes occupy important positions, e.g., as bottle-necks which all paths connecting different parts of the network must cross. If someone controls (or is) a bottleneck, they can use that to extract benefits from other members of the network, or to impose their will on them in other ways. One thing which makes me increasingly uneasy about such reasoning, however, is that the existing network is not, usually, the only one which could exist. There are, as it were, other potential or counter-factual networks nearby. Surely, among strategic actors, the ability to exploit one's network position will depend on the characteristics of those potential networks and how easy they would be to bring into existence. If a node is very central in the existing graph, but it would be easy for its alters to switch their links elsewhere, it doesn't actually have a strong position, whereas a node which does not look very central, but whose partners have few or no other options, might be able to extract a lot more. I presume someone has formalized this game-theoretically. (I might even have read such a formalization and forgotten it.)

permanent link for this note   RSS feed for this note

Notebooks:   Powered by Blosxom