[Clyde Schechter]

I'm not sure what you're asking here. There is nothing in your -xtreg- analysis here that hinges on the data being quarterly. If it were the exact same data, but the dates were yearly, or daily, or whatever, you would get exactly the same results.

[Aliza Khan]

I am attaching the screen shot of data editor. I have checked the format for 'quarters', it is in quarterly format. But in the data editor the year ins not 2008, it shows as 2462 q 1.

Please help.

[Aliza Khan]

Let me rephrase my question. The data in quarter format is shown as Q1-2463, not Q1-2008. Why is stata reading it like that? When All I entered was Q1-2008? and then Q -2 2008?


Many thanks for your help and patience uptil now. I really appreciate that. I understand that how questions being asked by someone who doesnt know the software can be quite annoying. Thank you for your time and help.

Waiting your response.

[Clyde Schechter]

Well the screenshot was readable (many are not, which is one reason why screen shots are discouraged here.) Where and in what way did you "enter" these dates? The odd thing is that the numerical value underlying what is being shown as 2462q1 is, wait for it..., 2008! So somehow your 2008, etc. have been picked up by Stata, And then when format %tq is applied, 2008 is appropriately translated into the quarter that is 2008 quarters after the beginning of 1960, which is the first quarter of 2462. But I don't know how those numbers got in there in the first place: you speak of "entering" the data, and I don't know really what that means.

All I can say is that the numbers you have in there are not correct. Not knowing how they were created, I can't advise on how to fix the process.

[Aliza Khan]

I just imported the values from excel sheet. where in quarter column, I wrote quarter as '2008q1', which I guess is being read by Stata as "2008th quarter after beginning of 1960). How can I correct it so that it reads it as 1st quarter of 2008.

[Clyde Schechter]

I'm a bit baffled. How did you get the data from Excel to Stata. If you are using -import excel-, then 2008q1 should have come in as a string variable. Then you could have run

gen quarterly_date = quarterly(B, "YQ")
format quarterly_date %tq

and everything would have been fine.

But somehow, whatever process you used to import the data into Stata turned your 2008q1's into just 2008. That's how you ended up with those bizarre dates.

I suggest you go back to the import process. Check the spreadsheet to make sure that the dates really do look like "2008q1" etc., and that the q1 parts haven't somehow been lost. Then when you're sure of that, use -import excel- to bring in the data and do what I suggested above.

[Sillis Kojo]

Greetings everyone. I am a little new to stata and really want to know if there is any stata command to convert annual financial data for the period 2005-2017 into a quarterly data in stata 13.1. Any assistance will be greatly appreciated. Regards

[Clyde Schechter]

Speaking of "converting" annual into quarterly data is a loose way of thinking that will, sooner or later, get you in trouble.

This situation typically arises when people have two data sets they would like to link up, and one has annual observations but the other has quarterly observation. If the resulting linked data set is made to be annual, then some of the richness of the quarterly data is inevitably lost. If the resulting data set is made to be quarterly, then some kind of imputations must be made to fill in the unknown quarterly values of those variables that originate in the annual data. Any such imputations rely on assumptions about the evolution of the variables in time, and such assumptions range from reasonably well supported by empirical research to outlandish and completely ad hoc.

There are numerous ways of doing that, and the commonest ones certainly can be implemented by Stata commands. That's the least of your problems. But you need to have a plan that is based on an understanding of your data, what they mean, and what questions you want to answer with them. It sounds like your desire is to somehow impute quarterly values of variables within your years. How, mathematically do you plan to do that? There are simple ways of doing it, such as just taking the annual value and "spreading" it to all four quarters of that year. Another simple approach is linear interpolation. More complicated is the use of more complicated and more sophisticated imputation models that take into account the values of other variables available in the two data sets. The appropriate choice really depends on your research goals, and which method will best advance them. Previous research in your area may have grappled with these issues before and reached useful conclusions.

After you've decided what kind of imputation process you want to use, finding the appropriate Stata commands will be quite simple.

[Sillis Kojo]

Thanks very much Clyde. Very helpful. I would be very much grateful if I can get the code to linearly interpolate my annual data series into quarterly ones in stata. Best regards

[Clyde Schechter]

-help ipolate-

[Nick Cox]

I don't know what's envisaged here. If someone were forcing me to do this I would assign annual values to notional quarters 2.5 and then interpolate for quarters 1, 2, 3, 4. But if you are insistent on interpolation, why insist on linear interpolation?

[Sillis Kojo]

Thanks, Clyde. Well noted.

[Sillis Kojo]

@Nick. Actually, I am trying to do a factor augmented autoregressive time series analysis for some selected insurance firms with annual feasible data from 2007 to 2015 which gives me only 9 data points for each variable. the augmented factors will be principal components of the original variables to capture latent factors. I am only trying to increase the data points by increasing the frequency of the series which I think is inadequate for PCA. Your comments on time series analysis with nine observations will also be much appreciated