The academic way of thinking

random
Back

This morning when I was doing my game theory homework, I suddenly realized how the half-year training as a PhD student has changed my way of thinking about business issues.

When I was an undergrad, I frequently involved in different types of business dicussion: writing column articles for a business media, making stock pitches in an investment club etc. Thinking back on these discussion, the method I used is mainly common sense reasoning. I collected information, and tried to form a coherent narrative on a question.

To illustrate the economic moat of a firm, I would use survey on consumer satisfaction, description on supply chain speed etc. to support the point. I would read interviews of the CEO, try to grasp her vision and evaluate whether she is trustworthy or lot, largely based on qualitive judgement. To analyze competition, I would compare the products of different companies, try to summarize what are their respective value proposition, and try to predict who can win in the long run.

The academic way of thinking, in my understanding, is related, but different from the approach I used before. There are two components to it: theory and data-driven analysis.

Theory

First, let's talk about theory. My current conception about theory is that it is usually stemmed from a phenomena of interest, and then (1) provide an abstraction on the essential elements of the phemenoma (2) decribe the relationship between the elements and (3) provides meaningful prediction on the counterfactuals.

Cournot duoply game theoratic model, for example, focuses on the real-world phenomena where two firms are competiting with each other in an industry by setting different production quantities. The model specifies (1) the profit function of the two firms and (2) demand curve of the whole industry. In other words, these are the two elements the model thinks is crucial to determine the final market outcome. Using the concept of Nash Equilibrium, we can quantitatively determine what's the final production level of the two firms.

Compared to using common sense reasoning, this duoploy model forced me to pick the most essential elements of the competition, decribe them using formulas and use the power of math to discover the answer. I won't get lost in the complex institutional details or product feature, or suffer from the ambiguity of my prediction. The model provides me a tool to think about the issue with simplicity, precision and sharpness. Even though people criticize these model for being unrealistic, they provide at least a type of narrative that can be used to analyze an issue. At least to me, they are sometimes much more trustworthy than common sense reasoning.

In this regard, I'm very much exciting to taking industrial organization course next semester to grasp this way of thinking.

Data-driven analysis

The second power tool is data-driven analysis. In common sense reasoning, I usually put too much focus on one or two data points, relying on them to support an argument. But it's likely that they may only represent a very special type of scenerio and cannot be extended to a broader scale. Perfoming the analysis based on a large sample of observations can help alleviate this bias.

Even more fascinating is the more accurate causal inference given by the statistical methods. Causal inference using purely common sense reasoning is usually arbitrary and contains lots of noise. Big data set + statistical methods gives us a much stronger tool to discover the true causal relationships. I'm not saying that there's no limitation to the method. Jeff Bezos may know much better about how to success in the retail industry then the answer from an academic. But there are certain type of questions where academics can provide a much more trustworthy answer using scientific methods.

Conjoint analysis, for example, is a statistical method in marketing that can be used to evaluate how much customers value different features of a product. Firms can use this method to decide the optimal pricing strategy of a new product. Using purely common sense reasoning, the firm can only use simpler methods such as benchmarking with competitors, cost-based pricing etc., which cannot taken into account the important granular information of customer preferences on different features.

Concluding thoughts

Theory and data analysis are truly fascinating to me as two new tools to think about business issues, or the broader human history. But still, I think the highest level of original thinking is still common sense reasoning. The great thinkers in the history, Aristotle, Plato, Confucious, Nietzsche etc. never learnt social science theory or data analysis... Their ideas are original, describing highly abstract subtleness of individuals and society that cannot be summarized using math. When I am taking a deep dive into academic theory and data-analysis technique, it's important to keep in mind that bringing back to common sense reasoning may still be the key to real original thinking.

© Zhiwei (Berry) Wang.RSS