Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Error with ffirst in ivreg2

From   Abigail Wozniak <>
To   "" <>
Subject   RE: st: Error with ffirst in ivreg2
Date   Wed, 6 Apr 2011 11:44:00 -0400

Thanks, Kit. I tried running with the first option instead of ffirst and was getting the same error. But I think I diagnosed the problem. 

The first and ffirst options were not compatible with the D, L, and F operators in my specifications, although the ivreg2 specifications using the L/D/F operators ran without these options. When I substituted "hardcoded" variables equal to the values of the lead/lag/change, the ivreg2 spec with the ffirst or first options works.

Does anyone know why this is? I know some commands do not allow time series operators, but it still seems odd to me that ivreg2 can use TS operators without the first/ffirst commands but not with. I would have expected it to fail in all cases or not at all.

Abigail Wozniak
Assistant Professor of Economics
University of Notre Dame
Tel: 574.631.6208

-----Original Message-----
From: [] On Behalf Of Christopher Baum
Sent: Wednesday, April 06, 2011 7:16 AM
Subject: Re: st: Error with ffirst in ivreg2

Please show us your logfile for that estimation, w/w/o ffirst option. Does the first option have the same problem?


On Apr 6, 2011, at 2:33 AM, Abbie wrote:

> I am running a 2SLS specification with two endogenous variables and two instruments, plus state and year fixed effects.  One endogenous variable is the current period population and the other is the one year lagged population. The instruments are the IV for the current period and the previous period's instrument.
> The first stage with one endogenous variable is strong, but I wanted to check that both instruments are strong predictors in the two instrument-two endogenous variable case.
> When I run my specification using ivreg2 with the ffirst option, I get the error message "reduced form estimation failed." However, when I omit the ffirst option, the same specification runs and I get reasonable looking estimates.
> Does anyone know what the problem is? It seems odd that ivreg2 would generate 2SLS estimates if the reduced form truly were not available so I don't understand why it runs without ffirst but not with it.

Kit Baum   |   Boston College Economics & DIW Berlin   |
                              An Introduction to Stata Programming  |
   An Introduction to Modern Econometrics Using Stata  |

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index