Missing value found in signal transition matrix?

[bluelines2]

Hi all,

I am trying to replicate a paper of Diebold et al: "The macroeconomy and the yield curve: a dynamic latent factor approach (2006)"
I used exactly same data as the paper used and I confirmed several times that there is no problem on my cording for the state-space system.
But when I hit estimation, an error massage pops up saying "Missing value found in signal transition matrix"
I could not understand what the missing value is indicated.
I attach my work file here since I have to solve this problem asap.
Please help me and see my yields-only model.
Thanks

[EViews Glenn]

If you take you sspace object and select View/Specification/Coefficient Values you will see that the coefficients for SV2 and SV3 in the signal equation are indeed NA (missing) values as indicated by the error message. The Kalman filter algorithm cannot be run when there are missing values in the structural equation coefficient values.

If you now select View/Specification/Coefficient Descriptions and look at the expressions used to obtain the coefficient values, you'll see that the SV2 and SV3 coefficients are expressions which have the coefficient C(36) in the denominator of a fraction. Looking at the current value of C(36), we see that the current value is 0. Hence the missing values.

Bottom line is that you are going to have change the starting value of C(36) to something nonzero. If I were you I would also tighten up my convergence value. The default value of .001 is probably a bit loose. I'd go to at least 1e-6 or so.