Forecasting estimation with panel data.

[Sean Bilic]

I try to do forecasting estimation with panel data. I want to estimate GINI under the impact of covid 19 pandemics.

For that, I regress GINI coefficient of GDP, Health Expenditure and WPDI (= world pandemics discussion index).

My code:

forecast create, replace

xtreg GINI WPDI log_GDP log_HE, re

estimates store myforecast

forecast estimates myforecast

forecast solve, prefix(f_) begin(2010) end(2021)



FIRST PROBLEM: my data set is between 2010 and 2021. And I want to estimate the period between 2022 and 2034, (i.e., after 2022). However, this estimation gives result in the period 2010-21. Do not allow for expanding the time span.

For example when I use the command

forecast solve, prefix(f_) begin(2022) end(2034)

This command is not allowed.



SECOND PROBLEM: the results forecasted seem not to be accurate and correct. all forecasted results of GINI are obtained to be equal to approximately 31. However, there are many countries having gini coefficient fo 45 or above in the dataset.

How can run forecasting estimation correctly?

Thank you.

[Jared Greathouse]

Using Statalist is just like making wild mushroom risotto. If I asked you to make my favorite version of risotto, you'd likely need to know what the SPECIFIC ingredients are and the steps you do to make said dish. So far you've given the recipe (the code, impropriety formatted), but there's no ingredients (the data).

Precisely, you must give your dataset using dataex and your code that you've tried (formatted accordingly). No dataset, no code=no real assistance.


I'm not trying to be mean in asking this, but let's say I walked up to you on the street and showed you this question. Would you be able to diagnose the problem short of having a dataset to try this with? No, you wouldn't. When I ask questions, I either give my real data or I give some of my example data. Sean Bilic

[Jared Greathouse]

I want to estimate GINI under the impact of covid 19 pandemics.

How? Everywhere was affected by the pandemic, so presumably there's some specific intervention or variable you're interested in. What is

WPDI (= world pandemics discussion index).

? And also, why do you want to use the forecasting command? I'm not saying it's illegal, I've just never met anyone who's needed it.

[Sean Bilic]

Using Statalist is just like making wild mushroom risotto. If I asked you to make my favorite version of risotto, you'd likely need to know what the SPECIFIC ingredients are and the steps you do to make said dish. So far you've given the recipe (the code, impropriety formatted), but there's no ingredients (the data).

Precisely, you must give your dataset using dataex and your code that you've tried (formatted accordingly). No dataset, no code=no real assistance.


I'm not trying to be mean in asking this, but let's say I walked up to you on the street and showed you this question. Would you be able to diagnose the problem short of having a dataset to try this with? No, you wouldn't. When I ask questions, I either give my real data or I give some of my example data. Sean Bilic

I construct dataset by myself by using WORLD BANK database website. How can I upload my data file here?. Also, I write my code.

[Jared Greathouse]

I know, and I like that you've written you code, but we need both, the data, and the code. Here's my advice to you: if you're going to be on Statalist for the long term, then watching this is of interest to you. Of course, we explain this in the FAQ, but what you'll need to learn to do is present data examples by using dataex. The YouTube video (the first link) shows you how.


Put simply, while we recognize explanations of data are well intentioned, say

FIRST PROBLEM: my data set is between 2010 and 2021. And I want to estimate the period between 2022 and 2034, (i.e., after 2022). However, this estimation gives result in the period 2010-21. Do not allow for expanding the time span.

For example when I use the command

forecast solve, prefix(f_) begin(2022) end(2034)

This command is not allowed.

this doesn't tell me much of anything about what the real problem is. You don't tell what the error is, and I can't even try to reproduce the error because you've not provided an adequate dataset. I'm sure there's an answer to this question by the way, but what you'll need to provide at a minimum is a data example. The YouTube link will explain this.