Partial Identification of Discrete Instrumental Variable Models using Shape Restrictions

• DOI: 10.2139/ssrn.3711861
• Publication Date: 2020-12-03T11:50:36Z
• Authors: Takuya Ishihara
• Keywords: Instrumental variable; Identification

This study examines the nonparametric instrumental variable model with discrete instruments and explores partial identification estimation of target parameter, which is a linear functional structural function. We include numerous parameters, such as difference between values function at two different points average effect hypothetical policy change. Informative bounds on parameter are derived using control approach shape restrictions. Illustrative examples demonstrate that restrictions have power. The lower upper estimated sieve method we show our estimator computationally convenient consistent. An empirical application illustrates usefulness method.