[Kaushik Borah]

I have panel data of agricultural plots for 3 years. The main reasons for the unbalancedness come from crop shift and crop failure. So, if there is crop failure/shift, the database does not record any of the variables. However, I have at least one year of data for each plot. So, if data is missing for any year then the whole set of variable is missing.

Jeff Wooldridge Here can I use xttobit with time averages and complete case dummies? or Should I run CRE Tobit by keeping only balanced cases?

[Jeff Wooldridge]

I would try it with the balanced subsample but also create dummy variables for having only one year or two years available -- regardless of the pattern. The base case is three complete years of data. Then, when you include the time averages obtained from the complete cases, include the two dummy variables, say, tobs1 and tobs2. You might interact this we the time averages and maybe even the covariates in the orginal model. Just do something flexible, and then you can compare the APEs with the balanced case.

[Ange Kakpo]

Dear Professor @Jeff Wooldridge. I am experiencing a similar issue to @Olive Bat. I have panel data with N=3000 (households) and T=2 (2011 and 2014) and I want to estimate a Tobit model with households fixed effects. I would like to know if it is correct to do: tobit y x i.hid, ll(0) vce(cluster hid) ? Otherwise, what would you suggest I do? Thanks.

[Abdramane Cherif]

Hello,
I face a problem in understanding the structure of my data and in estimating. For my study, I use the World Bank Enterprise Survey conducted in more than 25 countries between 2010 and 2018 (e.g. Ghana 2010, Keyna 2010, Central African Republic 2011, Ethiopia 2011, Cameroon 2016, Chad 2018, Gambia 2018 etc.). In these countries, the survey was conducted only once, so there was no repetition in the observations.
The objective of my study is to assess the effect of corruption on access to credit. My dependent variable is a binary variable. I want to estimate a simple logit like in the article I read. To answer this research question, can I declare my data set as panel data or cross-sectional data. If it is panel data, that is fine. But if it is cross-sectional, then how can I process it in STATA to calculate by combining all countries together and how do I control for country effect, industry effect and year effect?
What is the Stata command?

Your suggestions are very much appreciated.
Thanks

[Carlo Lazzaro]

Abdramane:
why not providing an example/excerpt of your data via -dataex- (as per FAQ)? Thanks.

[Masanori Matsuura]

Jeff Wooldridge when a variable is an index, it's from zero to one, and around one-third of observations is zero, should I use the Tobit model or still use OLS?

[amira elshal]

Jeff Wooldridge Dear Professor Wooldridge. I am trying to estimate the effects of different firm determinants on the firm likelihood of paying informal payments. My dependent variable is informal payments as a % of sales, ranging from zero (where a firm does not pay any informal payments) to 100%. I have an unbalanced panel of firms (2013 to 2020, but with gaps). I would like to account for firm fixed effects. Which of the models below, if any, do you advise me to use, please?

xtreg inf_pay_sales x1 x2 x3 i.x4 i.x5 i.year, fe vce(robust)

pantob inf_pay_sales x1 x2 x3 x4 x5 panelid

xtpoisson inf_pay_sales x1 x2 x3 i.x4 i.x5 i.year, fe vce(robust)

xttobit inf_pay_sales x1 x2 x3 i.x4 i.x5 i.year

My concern is that my dependent variable (informal payments as a % of sales) is conditional on the firm's decision to pay/not to pay. We only observe this % when the firm decides to pay. Do we have to take into account the potential amount of informal payments that non-paying firms would pay IF they would decide to pay. This value is unobservable so I am thinking it has to be estimated using the Tobit model?

[amira elshal]

Dear Professor Jeff Wooldridge. Following my previous message, I forgot to mention that the number of observations is relatively big (but the panel is unbalanced). Please see below for using pantob.

[amira elshal]

Dear Professor Jeff Wooldridge. Please let me know if you require any further info. Thank you.