In This Set of Notes We’re Going to
Causality is a relatively simple concept that that can be conceptually quite interesting. At a high level, we all intuitively understand what it means for something to cause something else. You walk into a room at home, you flick a switch and the overhead light turns on. You hit the switch again back to its initial position, and the light goes off. The switch causes the light to turn on and off.
Many of the questions that we’re interested in practice are also causal in nature. For example, an interesting trend in the Housing Market, as Conor Sen from Bloomberg highlights this week) is that the age of the median first-time home buyer has been increasing.
One initial reaction is to ask why? What is causing this increase? Is it something about mortgage rates, demographics, bank lending that’s driving this pattern?
https://www.bloomberg.com/opinion/articles/2024-11-20/to-get-the-housing-market-moving-raise-property-taxes?srnd=undefined
The “why” questions are fascinating. We see something in the data and we want to know what’s the underlying reason. In statistics, econometrics, business analytics, we refer to these types of questions as reverse causal questions. They are questions about “what caused Y?”. Your high school history class had a bunch of these types of questions — what were the causes of WW1?
In analytics classes, we can partially answer (no not the Python built-in function! but good thinking) these reverse causal questions by focusing on a specific factor $X$ and trying to understand what impact it had on $Y$. For example, motivated by the above figure, we may try to understand what impact raising property taxes has on the age of age distribution of homeowners as done in a recent paper from the Minneapolis Fed.
By the end of this note, you’ll understand what the following summary of the paper means and hopefully you’ll have a sense of how “credible” such an analysis is (certainly a high bar, but I think one well worth striving for!).
<aside>
“Controlling for factors like house characteristics and location, the economists find that a doubling of property tax rates is associated with a 20 percent drop in housing prices and lower home-price-to-rent ratios. They further find that higher property tax rates are associated with a younger-skewing population.”
</aside>
To keep things concrete, we’ll consider how we might go about answering the following causal question — let’s say a business school, like Questrom, wants to understand the impact that taking a Python analytics course has on future earnings.