Dimensionality reduction is a topic that has governed our (the 2017 BGSE Data Science cohort) last three months. At the heart of topics such as penalized likelihood estimation (Lasso, Ridge, Elastic Net, etc.), principal component analysis and best subset selection lies the fundamental trade-off between complexity, generalizability and computational feasibility.
While deep methodological differences exist across economists, many disagreements involve “talking past each other.” Each side uses similar words to discuss fundamentally distinct, though related, concepts. This is especially a problem with every-day language words and leads to more confusion than understanding.
One problematic term is information. Everyone believes they have a reasonable definition and that others have the same concept in mind. This is unfortunate and stagnated the discussion. Only through clarity of thought and language can these issues be resolved.
Complete and Perfect Information, or Ignore for Now
Doing what was necessary for early models, the economists started easy. They ignored it. They approximated that every actor knows everything. That made life easy.
Since Marshall and Walras, economics focused on equilibria. Starting from perfect competition, complete and perfect information are crucial. How do supply and demand equilibrate? Everyone knows everything. After a few easy steps, boom, supply=demand.
While all economists admit perfect information is an untrue assumption, it is still the default in many models.