How Much Should We Trust Linear Instrumental Variables Estimators? An Application to Family Size and Children's Education
Many empirical studies specify outcomes as a linear function of endogenous regressors when conducting instrumental variable (IV) estimation. We show that tests for treatment effects, selection bias, and treatment effect heterogeneity are biased if the true relationship is non-linear. These results motivate a re-examination of recent evidence suggesting no causal effect of family size on children's education. Following common practice, a linear IV estimator has been used, assuming constant marginal effects of additional children across family sizes. We find that the conclusion of no effect of family size is an artifact of the linear specification, which masks substantial marginal family size effects.
|Date of creation:||Nov 2009|
|Date of revision:|
|Contact details of provider:|| Postal: IZA, P.O. Box 7240, D-53072 Bonn, Germany|
Phone: +49 228 3894 223
Fax: +49 228 3894 180
Web page: http://www.iza.org
|Order Information:|| Postal: IZA, Margard Ody, P.O. Box 7240, D-53072 Bonn, Germany|
When requesting a correction, please mention this item's handle: RePEc:iza:izadps:dp4562. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Mark Fallak)
If references are entirely missing, you can add them using this form.