§1. In this appendix, Marshall begins by discussing the tension between specialisation and generalisation (I think he’s going to argue for the former, given that economics split off from political science early in his career, a split that I’ve lamented).* He begins with a caution against narrow and deep:
Specialists who never look beyond their own domain are apt to see things out of true proportion; much of the knowledge they get together is of comparatively little use; they work away at the details of old problems which have lost most of their significance and have been supplanted by new questions rising out of new points of view; and they fail to gain that large illumination which the progress of every science throws by comparison and analogy on those around it. Comte did good service therefore by insisting that the solidarity of social phenomena must render the work of exclusive specialists even more futile in social than in physical science. Mill conceding this continues:—”A person is not likely to be a good economist who is nothing else. Social phenomena acting and reacting on one another, they cannot rightly be understood apart; but this by no means proves that the material and industrial phenomena of society are not themselves susceptible of useful generalizations, but only that these generalizations must necessarily be relative to a given form of civilization and a given stage of social advancement
I agree wholeheartedly with these concerns (read this from 2016 and an improved version from 2018) and wrote this in my January newsletter:
The humanities (language, history, philosophy) illustrate the diversity of human existence just as the sciences (biology, physics, etc.) illustrate our similarities. This explains how scientists can collaborate and agree on the “big picture” while failing to see the point of humanities studies that don’t seem to draw any conclusions (and sometimes seem locked in eternal battles over the “right” element drawn from a pile of subjective perspectives)
§2. Marshall admires the utility of deductive mechanical reasoning in economics but cautions against excessive reliance on models untested by experience and intuition. Further, he notes that the human subjects of economics — unlike the atoms of chemistry — are actively changing their forms, functions and reactions while “under the microscope”, which makes accurate conclusions less likely.
§3. Marshall advises using both deductive (logical) and inductive (historical) methods to understand (looking back) and predict (looking forward). Given the impossibility of living life in parallel universes, we need to be cautious in drawing conclusions but hopeful in seeking explanations for observed patterns.
§4. Given Man’s tendency to see patterns everywhere (including where there are none), Marshall cautions against aggressive claims to insight in assembling “pertinent causes” for observed effects. He explains how both strategy and tactics are important in naval warfare but difficult for analysts to later recreate. What decisions were not made; what information was used in making decisions, what information was unknown to the actors but known to later historians? He warns economists trying to explain individual decisions and their aggregates.
§5. Intuition (Marshall calls this “mother-wit”) and technique are complements: Wisdom draws from experience; technique pushes one to think about potential situations beyond that experience. (I often get interesting insights by looking at the “off-diagonals” of 2×2 figures.) The aggregation of knowledge over time allows each generation of academics look yet further, standing on the shoulders of giants.
§6. Economics can explain a lot but the accuracy and value of its explanatory power drops as its area of study expands.
* I was wrong, as he doesn’t come out in favor of either view, unlike later economists (see footnote below…)
This post is part of a series in the Marshall 2020 Project, i.e., an excuse for me to read Alfred Marshall’s Principles of Economics (1890 first edition/1920 eighth edition), which dominated economic thinking until Van Neumann and Morgenstern’s Theory of Games and Economic Behaviour (1944) and Samuelson’s Foundations of Economic Analysis (1946) pivoted economics from institutional induction to mathematical deduction.
I would say economics is getting more and more engineering problems. Models and mathematical explanations are best described if economists focus on understanding human behaviour in the market or in various institutional arrangement.