Show the community: *Can Visits to Car Dealers Predict Relative Stock Price Performance of Automotive Manufacturers?*

Research
, ,
1

I was curious how predictable raw visits would be for relative stock price performance among automotive manufacturers. In this study, I looked at the stock price performance of GM, Ford, Honda, and Toyota and connected it with car dealership visits. To me, it was kind of interesting. For GM and Ford, the connection seems easy to see, but not so much for Honda and Toyota. Can Visits to Car Dealers Predict Relative Stock Price Performance of Automotive Manufacturers? | by Thomas Young | Medium

This topic was automatically generated from Slack. You can find the original thread here.

2

Thanks for sharing - enjoyed reading this!

Reminds me a lot of Hengda Jin, Stephen Stubben, and @Karen_Ton’s recent publication, Customer Loyalty and the Persistence of Revenues and Earnings. They used foot traffic to retail locations to measure customer loyalty and found that customer loyalty influenced stock returns in the subsequent quarter earnings announcement. If there’s others in the Community interested in checking out their paper, you can find it here.

Hey @Ji_Mi_Park - I wanted to flag this article by Thomas for you as you might find it helpful for your research. I know you wanted to also use our Patterns data to see how it influences earnings for different businesses. Might be worth connecting with Thomas for any tips as you embark on this new line of research!

Hey @Minjae_Koo - have you checked this article out? Be sure to drop any questions for Thomas in this thread!

3

This is super cool and easy to follow! I think it serves as a good introduction to how you might use SafeGraph information to predict returns. I did have a few questions / comments:

  1. I really appreciated the Stata examples on how to merge the data – I’m an R user, so I’m pretty helpless at joining data sets in Stata. You might also include the code for how you made the graphs at the end of the post; it looks like they were produced in Tableau rather than Stata?
  2. I think it might help to expand out your results paragraph to explain what an R^2 and p value mean. That is, you might say something like “we can explain 6% of the variation in a firm’s stock price just from seeing how often people visit the dealership. We also find the relationship between visits and stock price is highly significant; it is extremely unlikely that this relationship shows up in the data by chance”.
  3. Car companies are often large conglomerates; is it possible to determine how much of each company’s business is actually selling cars? Maybe Toyota’s stock was less correlated with visits to dealerships because more of its revenues comes from buses as opposed to cars. It might make sense to control for % of revenues from new car sales.