1: \begin{abstract}

2: 

3: 

4: 

5: This paper proposes confidence regions for the identified set in

6: conditional moment inequality models using Kolmogorov-Smirnov

7: statistics with a truncated inverse variance weighting with increasing

8: truncation points.  The new weighting differs from those proposed in the

9: literature in two important ways.

10: First,

11: confidence regions based on KS tests with the weighting function I

12: propose %

13: converge to the identified set at a faster rate than existing procedures

14: based on bounded weight functions in a broad class of models.

15: This

16: provides a theoretical justification for inverse variance

17: weighting in this context, and contrasts with analogous results for

18: conditional moment equalities in which optimal weighting only affects the

19: asymptotic variance.

20: Second, the new weighting changes the asymptotic

21: behavior, including the rate of convergence, of the KS statistic itself,

22: requiring a new asymptotic theory in choosing the critical value, which I

23: provide.

24: %

25: %

26: %

27: %

28: %

29: %

30: %

31: %

32: %

33: %

34: %

35: %

36: %

37: To make these comparisons,

38: %

39: I derive rates of convergence for the confidence regions I

40: propose along with new results for rates of convergence of existing

41: estimators under a general set of conditions.  A series of examples

42: illustrates the broad applicability of the conditions.  A monte carlo

43: study examines the finite sample behavior of the confidence regions.

44: \end{abstract}

45: