[Justin Matz]

What could be the id identifier in my case, given that I have only macroeconomic variables, returns, and dates?

[Joao Santos Silva]

Presumably a variable identifying the returns, but if you are estimating a model with fixed effects, you should know what the fixed effects are for, right?

[Justin Matz]

I don't know really how to handle all this... I am quite new on Stata.
I assume that my fixed effects are for my different countries. However I had create an "average" sheet fro those 13 different countries

[Justin Matz]

I think that I have solved my issue by implementing an "age variable" which count for the number of years, the ETF has, on average.

[Justin Matz]

I have the following error message : "option fe not allowed" when I type "xtqreg y x, fe"... How can I solve this issue?

[Joao Santos Silva]

Dear Justin Matz,

You do not need the option as the FE are always included. The ID indicator in your case should probably be the country identifier.

Best wishes,

Joao

[Justin Matz]

Dear professorJoao Santos Silva,

Yes, it is what I did. I have 13 sheets, one per country, with 80 variables and 166 observations per variable (166 months).
I have gathered the 13 sheets in one --> each sheet below the other, which makes a lot of lines. and put a country identifier for each country.
Does it seem correct?

Kind regards,
Matz Justin

[Joao Santos Silva]

As far as I understand, yes.

Best wishes,

Joao

[Justin Matz]

Thank you so much Joao Santos Silva

[Sarah Magd]

Hello Prof. Joao Santos Silva,
- My panel dataset has 12 years and 28 countries. Would that be suitable to estimate a quantile regression with fixed effects using the xtqreg?
- I am currently using the following command:
xtset iso_3 year
xtqreg l.energy l.GDP l.FDI, id(iso_3) q(0.25)
Is this the right code for my sample?

[Joao Santos Silva]

Dear Sarah Magd,

That is on the short side, but it should be OK. The command looks fine.

Best wishes,

Joao

[Sarah Magd]

Thanks Prof. Joao Santos Silva
I estimated my model and the coefficient of my main variable is insignificant in the scale function, but it is significant in the location function.
Would this mean that my main variable does not change over the quantiles?

[Joao Santos Silva]

Dear Sarah Magd,

If a variable has a zero coefficient in the scale function, its effects do not vary across quantiles. However, having a statistically non-significant coefficient does not imply that it is zero; it may just be that it is not estimated precisely enough due to the small sample.

Best wishes,

Joao

[Siddharth Kaushal]

Dear Professor Joao Santos Silva, I am trying to use Quantile Regression with Fixed Effects in my research. My dataset consists of an unbalanced panel dataset with 320 observations across 5 years. The distinct number of entities are 117 while the minimum number of entities for any particular year is 45. When I run the regression using xtqreg "WARNING: 15% of the fitted values of the scale function are not positive". Would you be able to suggest why this is happening? Is it because of the size of the panel dataset? Thank you for your guidance.

[Joao Santos Silva]

Dear Siddharth Kaushal,

I am afraid that with only 5 periods, you cannot reliably estimate a quantile regression with fixed effects. I suggest you try the "correlated random effects" approach.

Best wishes,

Joao