I encountered a paper by Vuolteenaho (2002) in which he uses pooled VAR model. I have some troubles understanding the idea. He uses firm level variables (log returns, ROE, etc.) and ultimately he obtains a single coefficient (table II on page 241) for a variable. I understand how a regular VAR works, but how is this one estimated (I assume that by using pooled OLS/WLS, as mentioned in the paper), but how can I write out this pooled VAR in equations (for example for two firms)?