Thread by @oliverbeige

I just learned that the distinction between industrial engineering and economics isn't all that obvious to outsiders, so here's a short summary thread.

TLDR: industrial engineers are to firms what industrial organization economists are to markets.

The biggest commonality is that both have a very technical view of their core subject, using formal models to describe and improve the overall functioning of it. The overarching goal is to improve efficiency and minimize loss.

[For those who find that too soulless, this also includes minimizing the loss of life. A massive impetus to go down the path of driving automation was to put an end to the 1 million casualties from vehicle accidents.]

So both fields have this habit of seeing their subject, firm or market, as a little sputtering machine that requires constant attention and fine-tuning to run smoothly. Industrial organization evolved specifically from the theory of imperfect competition.

There's an significant overlap in methods used, mostly deriving from linear (and the various nonlinear) optimization models. Industrial engineering has about 20 years of headstart on methods, largely bc it isn't constrained by linearity considerations.

It would be a bit limiting to restrict IE to "decision science under uncertainty", but decision theory and graph theory (incl networks) are important tools.

The most important tool of modern IO, game theory, could be described as "decision theory from multiple perspectives".

The biggest difference is in the tools used for empirical testing. IE is in this regard very much in the engineering tradition of building things, creating test data, and stress testing the system until failure points are known and fixed.

IO follows economics in the trend towards applying the scientific method, meaning econometrics, and in particular OLS regression analysis.

A big reason for that is that traditionally, economists worked for public organizations, where the burden of proof was inordinately higher.

Working on adjacent subjects (firms and markets) and using a similar toolset surprisingly doesn't mean there is a steady flow of exchange between the fields, as the glacial speed of the knowledge spillover of things like machine learning (IE»IO) or game theory (IO»IE) shows.

Another notable distinction is the preference for dynamic/longitudinal (IE) vs comparative-static (IO) analytical methods. This might be due to the fact that the firm is seen as an ongoing concern while the market facilitates spot transactions.

But it is also much easier to build longitudinal models when there is only a single actor for which things have to be optimized.

After some 20+ years of mostly radio silence, there's some first signals that the fields are again comparing notes — or at least I now see more young professionals with both "Econ" and "ML" in their twitter profiles.

Which will come handy as more economists are joining the ranks of tech companies, they're likely sitting next to each other on the workbench.

And we'll see more people switching back and forth...

@threadreaderapp unroll plz oldmate