Welcome to Catholiceconomist.com
Structural Vector Auto Regression and the Wholesale Price Per Gram of Cocaine
I have been working on another idea, a structural VAR model and the macroeconomic effects on the wholesale price per gram of cocaine. The data are from the DEA STRIDE from 1985 to 2009, prices per gram of cocaine are from undercover drug buys of four ounces or more. The exogenous variables are temperature from Colombia and the entire case count of the DEA of all substances. The Macroeconomic variables are the log difference of economic growth, the log difference of the Colombian exchange rate and the number of kilos sized by the DEA of trafficking quantities of cocaine (>500 grams). The constraints are that all of the variables can affect the price per gram and the economic growth rate in the United States can affect the Colombian exchange rate. After a headache getting my SVAR code to work I got the following Impulse response functions.



The increase in Kilos seized represents increasing supply conditions which lowers the price per gram as one would expect. Economic growth raises the price of cocaine and Colombian exchange rate depreciation raises the price of Cocaine possibly due to the effect it has on inflation in the country. I will open it up for comments below. I'm new at this but it was an idea that seemed to work so far. Now I have some more reading to do on SVARs.