dfuller - no observations r(2000) and gaps in time serie.

[Christian Brink]

Hi all,

I'm getting the error above, no observations r(2000) for the ADF test.

Maybe it is because I'm using hourly data, but haven't tsset it the right way?
I have data like this: 01jan2016 00:00:00 and the next being 01jan2016 01:00:00.

Furthermore, where can I see which data im missing?

I use the following commands:
gen double Dato_1 = clock(dato, "DMY hm")
format *1 %23.0f
format *1 %tc
tsset Dato_1
tsset

tsline spanien
dfuller spanien, regress lags(1) trend

Output for tsset:
time variable: Dato_1, 01jan2016 00:00:00 to 03dec2020 17:00:00
but with gaps
delta: .001 seconds


Hope it makes sense, thanks in advance.

Kind regards Christian

[Nick Cox]

The default is to take the time step for date-time variables as 1 ms, as tsset shows. You need to

tsset Dato_1, delta(1 hour)

as help tsset explains in detail.

[Christian Brink]

Litterally just what i needed, it works just fine now !!! You are the man.

Thank you so much !!!!

[Justin Matz]

Hello everyone,

I have the same kind of question...
I have the following code:
". tsset Date

Time variable: Date, 01jan2008 to 01nov2021, but with gaps
Delta: 1 day

. tsset

Time variable: Date, 01jan2008 to 01nov2021, but with gaps
Delta: 1 day

. dfuller Date
no observations

. dfuller CPI
no observations
r(2000);"

Whereas I do a regression, or even "tsline CPI" it works...

Could someone help me to fixe this issue please?
Thanks in advance

[Nick Cox]

#4 No data example -- see https://www.statalist.org/forums/help#stata for how to give us one -- but here's my guess, not least because I wouldn't expect a CPI to be issued daily.

You have monthly data that happen to be referred to daily dates for the first of each month. If you tell tsset they are daily data, then commands can't find e.g. the predecessor to 1 Feb 2008, which would be 31 Jan 2008, or the successor, and so on. So, simple operations like L1. or D1. all yield missings.

If so, the remedy is to recast to a monthly date and tsset in terms of that.

gen mdate = mofd(Date) 
format mdate %tm
tsset mdate

[Justin Matz]

Nick Cox Amazing! Thank you so much

[ella hopkins]

Hi everyone,
having pretty much the same issue as others have had with the Dickey-fuller test, and I'm very inexperienced with stata
I'm working with daily stock return data which has gaps due to the fact that stock markets are closed over the weekends, my do file currently looks like this
generate trend = _n
generate time = td(18feb2015) + _n-1
format time %td
generate time1 = date(date,"MDY")
format time1 %td
tsset time1, daily
output looks like this
Time variable: time1, 18feb2015 to 25sep2015, but with gaps
Delta: 1 day
when i try to carry out an ADF test on my variables it says
no observations
any help would be greatly appreciated