Duke Economics Working Paper #97-27
This paper calculates rates of convergence of sieve estimates for general objective functions and pseudometrics. The theory is general enough to handle most parametric and nonparametric settings. It also allows for different types of dependency assumptions. An important application is rates of convergence for series expansions. Two examples are worked out in detail: nonparametric and semiparametric index regression. The estimates are performed by a penalized spline and a polynomial series expansion.
JEL:
Retrieve document:
46 pages