Control time effects for quarterly data

[Louis Tran]

Hi everyone,

I'm a student and I am quite new to Stata. I'm doing my thesis and being stuck at this point. I hope that you can help me with your precious advice.

As you can see in the picture in the last column, I have a quarterly time series. I wonder how I can control time effects by using Stata with this kind of data.

I really appreciate your time and kindness.

[Carlo Lazzaro]

Louis:
short answer:

i.time

What I can't get from your post is whether you're going to do a panel data regression with a continuous regressand.
Is that were the case, -reshape- your data in -long- format and see -xtreg-.

[Louis Tran]

Louis:
short answer:

i.time

What I can't get from your post is whether you're going to do a panel data regression with a continuous regressand.
Is that were the case, -reshape- your data in -long- format and see -xtreg-.

Thank you for your reply

I think my data is not a panel data set. It's simply the quarterly data though. In this case, am i right that i only need to put i.time to my regress command. Like this:
"regress droa froa fshare gdp_gr i r agr cgr car i.time"

However, the result showed is very weird. there is collinearity problems and the r squared equals to 1.

Hope you can help

[Carlo Lazzaro]

Luis:
what if you omit -i.time- from the set of your predictors?
Another option: what if you investigate a possible turning point:

c.time##c.time

?

[Oscar Ozfidan]

The problem is the way you have coded your quarters. You need to create 4 variables (not 4 for each year) for quarters then do i.quarter instead of i.time

[Oscar Ozfidan]

recoding quarters then using i.quarter will take care of the seasonal effects. but if you want to also control or factor in the time trend just create a new variable 1-N (gen trend=_n) assuming there is a single observation for each quarter. if not you can still work with it but then you would be working with panel data

[Oscar Ozfidan]

Louis Tran What i meant by create 4 variables for each quarter was that you will have one variable named quarter. But that variable will have a unique value for the same quarter over time. For example, you can put 1 for Q1, 2 for Q2, 3, for Q3, and 4 for Q4. i.quarter will treat them as a category rather than as a numerical number. just to make sure my poor explanation does not lead you in the wrong path