Applied Information Economics

Definition

Applied Information Economics (AIE) is Hubbard's synthesized methodology, structured as five repeatable steps: (1) define the decision and its relevant uncertain variables, (2) quantify current uncertainty using calibrated estimates, (3) compute the value of additional information for each variable, (4) apply measurement methods to the highest-value variables, and (5) make the risk/return decision once no further measurement is economically justified — then return to step 1 for whatever follow-on decision arises. The distinguishing move is treating "what to measure" and "how much effort to spend measuring it" as themselves calculations, not judgment calls, by running steps 2–4 as a loop that terminates only when the expected value of more information drops below its cost.

In the Book

Chapter 4 first lays out the five steps as the organizing spine for the rest of the book; Chapter 14 shows how, after roughly 20 years and 80+ client projects, Hubbard's team operationalized them into four project phases — Phase 0 (expert identification and workshop planning), Phase 1 (decision modeling and calibration training, producing an initial Monte Carlo model), Phase 2 (value-of-information analysis to target measurement, iterating until no variable's information value clears its cost), and Phase 3 (final risk/return optimization and reporting). He grounds this in the EPA's Safe Drinking Water Information System (SDWIS) case: branch chief Jeff Bryan needed to justify funding but couldn't quantify "public health" benefits, so a 12-person, five-workshop AIE analysis discovered the real decision wasn't about SDWIS as a whole (never seriously in question) but about which of three specific upgrades — exception tracking, web-enabling, database modernization, totaling roughly $3.5 million — were individually justified. The chapter's other cases (Marine Corps fuel forecasting, ACORD data standards) follow the same pattern: a workshop-driven decision model feeding calibrated 90% CIs into a Monte Carlo simulation, then a value-of-information analysis that, in Hubbard's telling, almost always finds only one or two variables worth measuring further even when the initial model looked hopelessly uncertain.

Why It Matters

AIE packages the book's other concepts — calibration, Monte Carlo modeling, and expected value of information — into a terminable procedure: it tells you not just how to measure something but when to stop, which is the piece most ad hoc "let's gather more data" efforts lack. As a general pattern, it's a template for any resource-constrained inquiry (research, diligence, engineering validation) where the real question isn't "can we know more?" but "is knowing more, right now, worth what it costs?"