Saturday, August 8, 2020

Skyscrapers and alternative data - An example of correlation not prediction


Everyone wants to use alternative data. Use something different and create an edge. Some data are very interesting, tell a good story, and are predictive, yet novel does not mean better. Some original and eye-catching data may just be correlated with the some would regard as bland and mundane. The skyscraper index is a good simple example of alternative data.

The concept is easy and catchy. When there is a topping of a new tallest skyscraper, it is a sign that we are at an economic top and ready for a collapse. Construct the tallest building and rest assured that the economy is ready for a downturn.  

Look at today. We may not have had a new tallest building, but by the measure of super-tall skyscrapers in New York at the end of 2019, it seems like we were at an economic top. China, by the number of mega-skyscrapers competed in the last few years should also be ready for  downturn. Growth has had a for some time. I will note that after the fact, many can usually find a tall building that was completed that should have provided that elusive economic top signal.

A more careful analysis of the skyscraper index shows that it, at best, just correlates with the credit cycle. (See "Skyscraper Height and the Business Cycle: Separating Myth from Reality".) When there is cheap credit, big speculative buildings will be built. Given there is a lag between planning, construction, and completion, the tallest buildings may be finished when the credit cycle will turn. There is a lot of thinking with the skyscraper index for a phenomenon that derives from excess credit but may not be a leading predictor of economic doom. 

This may be same with a lot of alternative data. Data may be fascinating, catchy, and informative, but these alternatives may not be more predictive than other conventional variables. The standard for alternative data should be clear. It has to add more explanatory power, serve as an improved predictor, and be cheap to create. Still, I will follow the skyscraper index because it is great cocktail conversation.  




No comments: