# Young Economics

start

## Start

$e^{i\pi} + 1 = 0$

## Phenomena, Explanation and Prediction

• How to explain phenomena (y) around our human lives?
• In science, it is comparatively easy to draw an explaining variable (x) and the relationship between y and x (i.e. equations) because we can experimentally prove a treatment effect of x on y.
• In social science, whether in economics or business it is hard to find an experiment-like data. Rather, economists draws the mathematical equations from their own logical reasonings/thoughts such as 'optimized decisions made by utility maximizing agents' the equations arise from decision makings. Thereby they build an economic model to answer the what-if questions (i.e. policy implications). It is fairly obvious they are having hard time to narrow the gap between internal (i.e. theoretical) and external (i.e. empirical) consistency.
• In data science, supervised learning developers do put an emphasis neither on a causality nor on what-if questions. Their primary concern is prediction no matter how strange the included explaining variables are.
• So, what is your approach?
• NYU economist Xavier Gabaix on an alternative to the the rational-actor model: The most central open question in economic theory, as I see it, is how to model realistic economic agents. Traditionally, economists have relied on the rational-actor model, but it is clear that it is just a rough caricature. It has been greatly enriched by behavioral economics in the past 30 years. Still, we are far from a unified, versatile, believable alternative to the rational-actor model. I am hopeful, though, that this might be overcome—in part because of progress in the sister disciplines (psychology and neuroscience) and basic modeling, and also because empirical anomalies are forcing the economic profession to be more open-minded. Contributions by computer scientists and physicists will help inject new perspectives into economics.
• First, we will restrict the phenomena (y) to be explained.