data from the US Drug Enforcement Agency’s System To Retrieve Information on Drug Evidence

A few points about these pictures which I’ll be elaborating on in future posts:

• sub i, sub j: There is significant variation from city to city and presumably dealer to dealer or customer to customer, since they plot interquartile range.
• 3-D data: Since both purity and quantity affect the price, we’re really talking about a “price surface” — just like a volatility surface or the yield curve on Treasurys. And in fact there are even more dimensions to the data since it could be cut differently, and … well, I won’t say what makes for good coke.
• data collection: Do you really believe these numbers? Some undercover cop probably solicited drugs (I didn’t read the methodology section but just guessing). Does that seem like an error-free data collection process? But the same goes for macroeconomic data, financial data from companies, and so on. It comes from somewhere, it’s not “the truth” necessarily.