You have chosen to sponsor your bid up to a maximum amount of .
I would like a customized programming in STATA , a statistics software package. We work with STATA 11 & 12 using time series data and have a dataset of more than 400 variables. Our usual procedure is as follows:
We start by selecting one variable (1 out of 400) to be named as dependent. We start using the VAR command and run Vector Autoregression using only 2 variables each time (the dependant versus one drawn from the 399 of the dataset) with N lags (N<30). After every VAR, we perform the vargranger command and ckeck for Granger causality between the variable and our dependent.
We would like to have an automated system running the VAR, setting the number of lags, test the cranger causalities and deliver the result of the "successful" variabels that have granger effect on our dependent variable. So, in the end we can have a summary of the variables that are the best candidates in terms of Granger causality vis a vis a dependent.
In that way we can significantly reduce the time needed to perform 399 VARs every time we need to explore the relationships of a new dependent with the rest of the dataset.