Location: 368A Heady Hall
Description: Karim Chalak
“Some Results on Partial Identification and Measurement Error”
Abstract: The Gini-Frisch bounds partially identify the coefficient associated with an explanatory variable that suffers from classical measurement error, in a linear equation. This paper studies generalizing these benchmark bounds to the case of a nonparametric nonseparable equation. In particular, the paper provides a nonparametric analogue to the “forward” and “reverse” regression bounds. Moreover, in doing so, the paper provides a nonparametric analogue to the linear regression “attenuation bias” due to measurement error. By studying conditions under which the basic insights that underlie the Gini-Frisch bounds apply in a nonparametric nonseparable equation, the paper clarifies the scope of these quintessential bounds.