Tobit model using a panel data

[Ourega-Zoe Ejebu]

Hi there,
I am investigating the relationship between 'units of cheap alcohol' (dependent variable) and soci-economic regressors (income, occupation, age, gender, type of alcohol...).

Firstly, I used a Hausman-Taylor model and my results were ok.
Then I wanted to used a Tobit RE model (for comparison purposes) using the Stata command 'xttobit' as follows:


xttobit Weekly_cheap_units_new Low_income2 Intermediate_income2 Intermediate Routine_manual Total_number_of_adults Household_with_children5 ///
Gender Age3040 Age4050 Age5060 Age60p MeanWeeklyAlcoholLvl_0wks2 MeanWeeklyAlcoholLvl_0wks3 Year_of_purchase2 Year_of_purchase3 ///
Total_alcohol_spending Beer_lager2 Cider2 Fabs2 Other_wines2 Spirits2 Promotion_dummy Percentage0Week ///
if Social_Class!=6 ,nolog ll(0) vce(jacknife)


but I obtain the following message error:
flat or discontinous region encoutered.

I suppose I should simplify the model by removing some regressors, but all the variables are vital for the model and cannot be removed.

Is it ok to use a standard Tobit model (Stata command: tobit) instead since the panel is short (2008-2010)?

Thank you for your assistance.

[Carlo Lazzaro]

As per Stata 13.1 .pdf manual, entry -xttobit-, Example 1: did you try to increase the number of integration points in your model via - intpoints() option?
As far as your second question is concerned, my gut-feeling is that -tobit- would ignore tha panel structure of your dataset.

Kind regards,
Carlo

[Ourega-Zoe Ejebu]

Hi Carlo,
thank you very much for your suggestion.
After many trials, the additional command did not work; i.e. the model did not converge.

I read a paper (Non-linear model with panel data, Honore, No Date) suggesting panel could be ignored, using a pseudo-maximum likelihood providing the model corrects the standard errors for the fact that the observations are not independent. So, I am considering using Tobit with the Stata option 'vce(robust)'. What are your thoughts about that?

Plus, I am only using Tobit since I need the predicted probability of the dependent variable since this command doesn't seem to be available with the Hausman-Taylor (which is the preferred model) The stata command is as follows:
margins if Low_income2==1 & Mean_weekly_alcohol_level2==0, predict(pr(0,.))

Thank you for your assistance.
Zoe

[Carlo Lazzaro]

Hi Zoe:
as from -xttobit- entry in Stata 13.1 .pdf manual, the panel estimator is not different from the pooled (tobit) estimator when sigma_ui is unimportant (i.e., when rho approaches or equals 0). In your case, this hypothesis is difficult to test, since your model does not converge under -xttobit-.
Nor the paper you mentioned (downloadable from http://people.stern.nyu.edu/wgreene/...e-cwp1302.pdf; posters are requested to provide full reference, please) seems (as it probably should be) more theoretical than practical in this respect.
You should probably rely on the results of Hausman-Taylor model.
Eventually, -margins- is available in - xthtaylor postestimation- as well.
Kind regards,
Carlo

[Ourega-Zoe Ejebu]

Hi Carlo,
thank you very much for your comments.

Margins after Hausman-Taylor are indeed available. Unfortunately, the option 'predict' is not authorised (as Stata states) after a Hausman-Taylor model.
This is why I was keen on using Tobit with the 'vce' option, just for the sake of obtaining predicted probabilities.
Plus comparing standard 'margins' (i.e. margins if Low_income2==1 & Mean_weekly_alcohol_level2==0) after Tobit and Hausman-Taylor respectively, yield very close results.

Thank you for your help.
Kind regards,

Zoe

[TaniaRA_BIS]

Hi, I am trying to estimate a Tobit model with a panel dataset. I would like to test for hesteroschedasticity and serial correlation. Does anyone know
what would be the appropriate tests and the corresponding commands?
Many thanks and kind regards,
Tania

[Zainab Mariam]

Dear Professor,

I am using Stata 14. The data type of my research is panel data (unbalanced), the time period is 22 years, 5084 firms. My model includes 10 explanatory variables (x1, x2, …, x10) where 9 of these 10 explanatory variables are lagged one year, while only one explanatory variable is in the current time/present period (i.e., not lagged. It is in time t). In other words, the nine explanatory variables (x1, x2, x3, x4, x5, x6, x7, x8, x9) are lagged one year, while the explanatory variable x10 is in the current time.

The dependent variable y of my research is a limited dependent variable and truncated between zero and one. Thus, I applied the Tobit model by typing in Stata the following:

xttobit y L.(x1 x2 x3 x4 x5 x6 x7 x8 x9) x10, ll(0) ul(1)

The explanatory variable x1 is endogenous. Thus, I will use the Instrumental Variable Tobit (IVTobit) method. Accordingly, the first issue is about the static model and my first question is: can I use the command ‘ivtobit’ for panel data i.e., without the prefix ‘xt’? Is there ‘xtivtobit’ command to deal with a panel data model with a limited dependent variable and with the endogenous explanatory variable x1? If no, I kindly ask you please to guide me on what I should do.

The second issue is regarding the dynamic model i.e., the model includes the lagged dependent variable y as a regressor. The dependent variable y of my research is a limited dependent variable and truncated between zero and one. Thus, my second question is: what should I do in this case? how to apply the Instrumental Variable Tobit (IVTobit) method in dynamic panel data (the lagged dependent variable y is an explanatory variable) and the dependent variable y is a limited dependent variable and truncated between zero and one? What is the command?

The third issue is regarding having two endogenous variables in the dynamic model i.e., the lagged dependent variable y is endogenous regressor and the explanatory variable x1 is endogenous. Thus, my third question is: what should I do in this case? how to apply the Instrumental Variable Tobit (IVTobit) method in dynamic panel data (the lagged dependent variable y is an explanatory variable) and the dependent variable y is a limited dependent variable and truncated between zero and one? What is the command?

My fourth question is: do I have to use internal or external instrumental variables? can I use lagged values of the endogenous variable as an instrumental variable?

Thank you in advance.