SEM - LR test

[Iva Hasikova]

Hello everyone,
I am estimating this model in STATA

sem (VATgap -> vatrevenue, ) (cpi -> VATgap, ) (vatquota -> VATgap, ) (taxquota -> VATgap, ) (gge -> VATgap, ) (consum -> VATgap, ) (gcf -> VATgap, ) (openess -> VATgap, ), nolog iterate(500) nomeans latent(VATgap) nocapslatent

Instead of the result of LR test, I am getting only this warning: The LR test of model vs. saturated is not reported because the fitted model is not full rank. There appears to be 5 more fitted parametr then the data can support.

Does anyone know, where exactly is the problem? And how can I fix this?

Thank you very much.

Iva

[Joseph Coveney]

Does anyone know, where exactly is the problem?

Your latent factor has only a single indicator variable.

And how can I fix this?

I'm not sure what you're trying to do, but given that your latent factor is redundant in what you've shown, maybe try something like the following.

regress vatrevenue c.(cpi vatquota taxquota gge consum gcf openess)

Either that or reverse the arrows.

[Iva Hasikova]

Now I see, that my previous code was wrong, its bolded part was missing.
This is "correct" code: sem (VATgap -> vatrevenue, ) (VATgap -> gdppc, ) (cpi -> VATgap, ) (vatquota -> VATgap, ) (taxquota -> VATgap, ) (gge -> VATgap, ) (consum -> VATgap, ) (gcf -> VATgap, ) (openess -> VATgap, ), nolog iterate(500) nomeans latent(VATgap) nocapslatent

I am trying to estimate MIMIC model with one latent variable: VAT gap, two indicators: vat revenue and gdp per capita (gdppc) and several causes: vat quota, tax quota, general government expenditure (gge), consumption, gross fixed capital (gcf), openess of economy.

Thank you very much for your help.