EMPIRICAL POWER COMPARISON OF NON-NESTED TESTS FOR THE EVM: SOME MONTE CARLO EVIDENCE

Recently, Bodla and Bhatti (2007) revisited Davidson and MacKinnon’s (2002) well-known J test and noted that thought the test is simple to compute but lack small sample exact test computation properties. This paper is one of the attempts to compute a new version of the J test and compare its power performance with the various existing tests to see the relative strength of our test to be called as an approximately most powerful test. The main objective of this paper is to study Monte Carlo evidence on finite sample performance of the now modified non-nested tests of mismeasured regression models in EVM, Errors in Variables Models, setting to see if the power performance of the new test.