Question

Comment on the following statement: “It is always better to use as many instruments as possible in 2SLS. This allows to have the highest possible prediction accuracy of the endogenous variable” in the first step, and hence, provides gains in efficiency.

Answer 1

Two stage least squares is a technique is a method of regression used in the analysis of structural equation model.Its actually an extension of ordinary least squares method . It's used widely to estimate over determined model. 2SLS estimation involves the use of instrumental variables.

* However using as many instruments possible while estimation won't allow us to have accurate predictions of the endogenous variable. Such a practice would do more harm to our analysis than good.

> Adding too many instruments means that we are increasing the number of overidentifying restrictions.

> Increasing the number of overidentifying restrictions can cause severe biases in 2SLS.

>. The estimates will be biased even if there are thousands of observations.

> It means that estimate is too low or too high.

> There are high chances that Standard errors will be too small.

> Statistical tests won't be considered as significant.

> Predictions based on the model Wii be inaccurate .

Therefore its not advisable to use too many instrumental variables as it will affect our our regression results as well as forecasting ability.