Hey all,<br>&#39;<br>I&#39;m trying to master Johansen&#39;s test for cointegration, but I am a bit puzzled. When I ran a bivariate Johansen test on some data I&#39;m working on, the result was:<br><br><br>Case 3: Unrestricted constant<br>
Rank Eigenvalue Trace test p-value&nbsp;&nbsp; Lmax test&nbsp; p-value<br>&nbsp;&nbsp; 0&nbsp;&nbsp;&nbsp; 0,21652&nbsp;&nbsp;&nbsp;&nbsp; 20,783 [0,0062]&nbsp;&nbsp;&nbsp;&nbsp; 14,153 [0,0502]<br>&nbsp;&nbsp; 1&nbsp;&nbsp;&nbsp; 0,10802&nbsp;&nbsp;&nbsp;&nbsp; 6,6300 [0,0100]&nbsp;&nbsp;&nbsp;&nbsp; 6,6300 [0,0100]<br><br><br>
(Two time series of 60 observations each. ADF, PP and KPSS tests indicate that they both are I(1). )<br><br>As I&#39;ve understood it, this means that the test rejects both the &quot;no cointegrating equation&quot; and the &quot;one cointegrating equation&quot; hypotheses, but there can&#39;t be two cointegrating relations with only two variables, right? Could there be some sort of misspecification issue (e.g. wrong lag length) going on?<br>
<br>Thanks<br><br>/Olle<br>