Sven Schreiber a écrit : That's it. What surprised me is that I've got missing values (empty observations) even for the dummy variable instead of having 0 or 1. In addition, once I set the sample on dummy=0 and want to reset it on dummy=1 (using the --replace and --restrict flags) that's the full sample coming out. thanx, artur

On Sun, 13 Dec 2009, artur bala wrote: When you do smpl FRS=0 --restrict gretl is attempting to reconstitute a balanced panel for you, inserting "missing rows" if required. In this case it is struggling, because your restriction does not leave much of a panel, and it ends up being more confusing than helpful. I have now (in CVS) changed the behavior of smpl --restrict so that it doesn't strive to produce a balanced panel unless you give the new --balanced option. So, with your dataset and "smpl FRS=0 --restrict", you'd end up with a non-panel: then it's up to you to see if you can salvage a panel, if you want to. Allin Cottrell

Hi, It happens that I'm using gretl's french version and in the test calculator window the gnuplot command fails systematically for all tests. I checked all other languages and the gnuplot crashes only for one-variance test. best, artur

On Tue, 15 Dec 2009, artur bala wrote: 1) I'm assuming that if someone has a panel dataset, then they probably want to be able to apply panel-specific methods if at all possible. 2) In gretl, it is possible to apply such methods *only* if the dataset constitutes a balanced panel (with an equal number of time series observations for each cross-sectional unit). 3) However, the requirement at point 2 is in a sense "formal": some of the time-series observations may be composed of NA (missing) values. 4) Therefore, when a user sub-samples in a way that would "naturally" destroy the panel structure of the data, we offer the "--balanced" option as a means of reconstituting the panel structure. This necessarily involves "padding" the observations for least some of the cross-sectional units with missing values. 5) Note that such padding observations will not be used in calculating statistics of any sort. They are there purely to satisfy the formal requirement of balance (which we need for "accounting" purposes). In your example: ? smpl FRS=0 --restrict --replace --balanced Full data set: 686 observations Current sample: 1:1 - 97:3 (n = 291)  # there are 269 observations with FRS=0 That's right: the extra 291 - 269 = 22 'observations' are padding rows. But only the observations that actually have FRS=0 will be used by gretl in calculating any statistical results. If you don't want the padding, don't choose --balanced. Allin Cottrell

Hi, I am trying to run restrict command on time series from panel sub-sample, but I got the message "Can't do: the current data set is different from the one on which the reference model was estimated dataset is subsampled, model is not." However, I could get the result when I copy the sub-sample into new file although some missing values still exist. Is this the expected behavior? How to make restrict command works in this sub-sample? Thank you, Yuniarto Hadiwibowo <script> open paneldata #Panel# panel rd 0 rf restrict b2=1 end restrict #Individual# loop i=1..4 --quiet smpl code=$i --restrict --replace printf "\n\nCode $i\n" ols rd 0 rf --robust restrict b2=1 end restrict end loop </script> ----- Original Message ---- From: Sven Schreiber <svetosch(a)gmx.net&gt; To: Gretl list <gretl-users(a)lists.wfu.edu&gt; Sent: Thu, December 17, 2009 9:23:23 PM Subject: Re: [Gretl-users] sub-sampling panel data Allin, I'm not sure I understand... Allin Cottrell schrieb: ok, so it's not really a balanced panel then in the usual sense of the term, right? so that sounds to me like --balanced doesn't really give you a balanced panel, it "just" sets equal matrix dimensions. I would have expected that --balanced does something like finding the least common denominator (wich may involve discarding lots of observations of course). But I admit I haven't read the documentation... feel free to tell me that I should do that first. cheers, sven

Thank you for the prompt reply and fix.It works fine for ols now. I am sorry if I push it too much, but the same problem may still exist for the nls estimation. I understand that "restrict" command is still new to nls. <script> open greene14_1.gdt logs C Q scalar a = 1 scalar b = 1 scalar c = 1 scalar d = 1 diff l_C l_Q lags l_C l_Q loop i=1..6 --quiet smpl unit=i --restrict nls d_l_C = a + b*d_l_Q + c*(l_C_1 - d*l_Q_1) deriv a = 1 deriv b = d_l_Q deriv c = l_C_1 - d*l_Q_1 deriv d = -c*l_Q_1 end nls restrict b2 = 1 end restrict smpl --full endloop </script> Best regards, Yuniarto Hadiwibowo ----- Original Message ---- From: Allin Cottrell <cottrell(a)wfu.edu&gt; To: Gretl list <gretl-users(a)lists.wfu.edu&gt; Sent: Sat, December 19, 2009 2:31:59 AM Subject: Re: [Gretl-users] sub-sampling panel data to time series The problem was this: the fact that the dataset was sub-sampled was not being recorded on the model produced by the ols command. This was specific to models estimated within loops, and it's now fixed in CVS. There will be a new Windows snapshot with the fix applied tomorrow. Allin Cottrell