1: \begin{abstract}

2: This paper derives the rate of convergence and asymptotic distribution for

3: a class of Kolmogorov-Smirnov style test statistics for conditional

4: moment inequality models %

5: %

6: for

7: parameters on the boundary of the identified set

8: under general conditions.

9: %

10: %

11: In contrast to other moment inequality settings, the rate of

12: convergence is faster than root-$n$, and the asymptotic distribution

13: depends entirely on nonbinding moments.

14: The results require the development of new techniques that draw

15: a connection between moment selection, irregular identification, bandwidth

16: selection and nonstandard M-estimation.

17: %

18: %

19: Using these results, I propose tests that are more powerful than existing

20: approaches for choosing critical values for this test statistic.

21: I

22: quantify the power improvement by

23: showing that the new tests

24: can detect alternatives that converge to points on the identified set

25: at a faster rate than those detected by existing approaches.

26: A monte carlo study

27: confirms that the tests and the asymptotic approximations they use

28: perform well in finite samples.  %

29: %

30: %

31: %

32: %

33: In an application to a regression of prescription drug expenditures on

34: income with interval data from the Health and Retirement Study, confidence

35: regions based on the new tests are substantially tighter than those based

36: on existing methods.

37: 

38: 

39: \end{abstract}

40: