The ivreg2 package: How do I exclude the exogenous regressors as instruments?

[John Kim]

Hello, I'm using the ivreg2 package. The relevant syntax is,


ivreg2 depvar [varlist1] (varlist2=varlist_iv)


where varlist1 (e.g., W1, W2, W3) are the exogenous regressors or "included instruments", varlist_iv (e.g., Z1, Z2, Z3) are the exogenous variables excluded from the regression or "excluded instruments", and varlist2 (e.g., X1, X2, X3) the endogenous regressors that are being "instrumented".


If I understand, ivreg2 will

1. instrument X1 using W1, W2, W3, Z1, Z2, Z3

2. instrument X2 using W1, W2, W3, Z1, Z2, Z3
3. instrument X3 using W1, W2, W3, Z1, Z2, Z3


Can I make ivreg2 exclude the exogenous regressors as instruments? That is,

1. instrument X1 using Z1, Z2, Z3 only
2. instrument X2 using Z1, Z2, Z3 only
3. instrument X3 using Z1, Z2, Z3 only

Can I make ivreg2 use different sets of instruments for different endogenous regressors? For example,

1. instrument X1 using Z1 only
2. instrument X2 using Z2 only
3. instrument X3 using Z3 only?

I'd greatly appreciate any and all help in answering these questions. Thank you so much for your time.

Best,


John

[Phil Bromiley]

You didn't get quick response. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

I could be wrong, but I thought you needed to include the endogenous variables appearing in the main model as instruments to get a consistent estimator. I don't think the program will do what you intend. If you look around, it may be possible to find where someone on this list has discussed manually generating instruments in which case you could have full control.