Incorrect error message

[stoddj]

When I run model Mod_M22 in this file, it aborts with the message that there is missing data. But the monthly data set is complete from 1973 to 2019, which includes the period specified (2015-20190. The trouble seems to be with the scenario. If I run it without a Scenario, there is no problem.

[EViews Matt]

Hello,

EViews is correctly reporting that override data is missing in Scenario 2. That scenario excludes and overrides series DISPINCRP, but the series that should hold the override values, DISPINCRP_2M, is full of NAs. Perhaps that series didn't exist prior to your initial solution attempt and an empty series was created for you, but either way the override data is missing.

[stoddj]

Thanks so much, Matt. 2 further questions, please: 1) Where in the output or error messaging do you find this detail? Or is it from your knowledge of the program? 2) More importantly for me, in previous attempts, the *_2m data WAS generated by my scenario 2 override. But unexplained to me, this *_2m output was slightly different from the original dispincrp series that I had specified in the scenario. From my previous experience in using overrides in scenarios, that should not be the case.

Is it possible this problem is because dispincrp was itself a series generated by other output? And if so, will saving the new series as text and reintroducing it perhaps help? Thanks again for taking this up with me.

[random_user]

Maybe Matt has a better method, but logically, when you get an error like that in a model, it must mean one of two things:

1. There's an exogenous variable with no data over the sample period, or
2. One of your endogenous variables is a function of its own lag, and there is missing data in these lags. (for example, no data at t-1)

So, first you can check the exogenous variables, since that's the most likely culprit.
Open the model, click on "variables", and you'll see the variable "dispincrp" in red - that's a sign that it's overridden in that scenario (which means it's exogenized) ... and there are no values for it.

[EViews Matt]

The errors messages didn't directly implicate the variable, DISPINCRP, so a little investigation is necessary as random_user points out. The most significant clue comes from your description; the issue only arises in your custom scenario and not the default scenario. The model's scenario summary (View -> Scenario View) shows that only major difference between those two scenarios is the exclusion and overriding of variable DISPINCRP. A quick check of the override series, DISPINCRP_2M, reveals the troublesome NAs, mystery solved.

It's difficult to say how DISPINCRP_2M came to be full of NAs. In the workfile you shared, the only comment for that series indicates that it was created by the command "mod_pr120_2.solve", referring to a model object not present in the workfile. Regardless, perhaps a quick reminder of how excluded/overridden endogenous variables behave will help clarify things.

If a variable is...

• NOT excluded, then the variable's solution series, e.g. DISPINCRP_2M, will be computed from the variable's equation as usual.
• excluded and NOT overridden, then by default the variable's solution series will be copied from the variable's actual series, e.g. DISPINCRP.
• exclude and overridden, then the variable's solution series will not be altered.

Note that if an endogenous variable is overridden but NOT excluded, the override has no effect.

[stoddj]

OK, thanks Matt & Random User. You've given me things to investigate. I think I will start by deleting all the DISPINRPI variables, and generating the original DISPINCRP variable anew, so it is cleanly created within this workfile. Thanks for your expertise.
// Jim Stodder

[stoddj]

The crucial thing turned out to be to exclude but not to override -- so that the original variable was preserved. Thanks again.