https://padlet.com/superpronker/hz308o5bj6fs Write away en-us 2016-10-11 06:06:26 UTC 2020-12-17 11:39:22 UTC hello@padlet.com https://padlet.com/superpronker/hz308o5bj6fs/wish/855284954 I generally think it worked fine with the questions, which guide you through the feedback and ensures that everything is covered. However, I was missing a box for both problem 1 and 2 for additional thoughts and comments. I wrote a lot of stuff which didn’t necessarily fit in the boxes where I wrote them, simply because there was no where else to write it. 

Further, I don’t like that I am able to see who gave me feedback or for the people o gave feedback to know that I did it. I feel a need to be kinder and scale back on criticism, especially when I know the person whom I gave feedback. I think it restricts the amount of constructive criticism which is given, when you have to provide feedback for a friend. ]]> 2020-10-23 09:16:40 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/855284954 https://padlet.com/superpronker/hz308o5bj6fs/wish/855315955 I like that you give some guide lines for what would be relevant to give feedback on. Further, I don't like that people I give feedback knows it's me giving them feedback. ]]> 2020-10-23 09:44:20 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/855315955 https://padlet.com/superpronker/hz308o5bj6fs/wish/863813216 Good:
- guidelines for feedback

Improvement:
- anonymous 
-> (although I think it incentivises to be friendly and constructive, being extra nice also takes up way more time

- Guidelines for feedback could be improved 
-> ( can't think of something specific though)]]> 2020-10-26 22:47:49 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/863813216 https://padlet.com/superpronker/hz308o5bj6fs/wish/882358427 Our tobit estimation results are extremely sensitive to the starting value when estimating for the whole sample or different continents. What are the possible reasons for this? Is there anyway it can be solved?

]]> 2020-11-02 14:15:46 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/882358427 https://padlet.com/superpronker/hz308o5bj6fs/wish/993369658 I'm kind of worried that the exam will be as extensive and challenging as the projects during the semester. Those will be some hard 48 hours as my group has put what feels like 48 working hours into the projects. My impression is we are not alone]]> 2020-12-07 09:25:34 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/993369658 https://padlet.com/superpronker/hz308o5bj6fs/wish/993396108 When preparing our projects for the exam, would it be considered beneficial to incorporate insigts from the course gained after writing a given project, or is this just a waste of time? An example could be to use LASSO for variable selection in a project written before the lectures on the LASSO method.]]> 2020-12-07 09:37:57 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/993396108 https://padlet.com/superpronker/hz308o5bj6fs/wish/994116892 Could you provide some details of how you expect inference in the logit model to be carried out, both in regards of the estimated parameter values and especially in the elasticities? 
I have now given feedback to 2 groups, and all of us have different answers to some of the theoretical exercises, maybe you could somehow elaborate the type of model a bit? ]]> 2020-12-07 14:25:02 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/994116892 https://padlet.com/superpronker/hz308o5bj6fs/wish/995461774 (1) Our Projects now include our names to make it possible to reference them. Is it fine if we upload the chosen project and the exam project with our names or do we have to make them somehow anonymous for the exam?

(2) How exactly shall we upload all the files for the exam? We thought about putting all code files into two ZIP folders and upload the two pdf files separately. ]]> 2020-12-07 18:44:40 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/995461774 https://padlet.com/superpronker/hz308o5bj6fs/wish/997174358 Throughout the course I have been a bit uncertain about what to do when we get "local minimum possible". Do we just accept that or? Often I have also gotten very large t-statistics using fminunc. Mostly since the outerproduct is near 0 for some of the estimated parameters. Is that fine or is there any way to go around this.]]> 2020-12-08 07:20:14 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997174358 https://padlet.com/superpronker/hz308o5bj6fs/wish/997179472 Do we have to write on the front page which parts of the exam each group member has done?]]> 2020-12-08 07:22:55 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997179472 https://padlet.com/superpronker/hz308o5bj6fs/wish/997206003 We were wondering how to treat marginal effects of a tobit model with a modified error term distribution. For example the CLAD model. Wouldn't we need to assume a normal distribution of the error term either way to use the derived expressions of the partial derivatives from the lecture and exercise? Or is it a better idea to derive the partial derivatives for the specific version of the tobit ? In practice the partial derivatives can be quite difficult for more complex versions of the model and it was unclear to us how the ME of the CLAD model should be derived. ]]> 2020-12-08 07:35:59 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997206003 https://padlet.com/superpronker/hz308o5bj6fs/wish/997232440 Is there a reason to use robust SE for a random effects model. To us it seems that RE is inefficient under heteroskedasticity. ]]> 2020-12-08 07:47:58 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997232440 https://padlet.com/superpronker/hz308o5bj6fs/wish/997233764 I have used latex to write the assignments. It is pretty hard to get precise measurement of how many pages you have written. How should we measure this. By counting the characters and if so are there any bandwith on that?]]> 2020-12-08 07:48:38 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997233764 https://padlet.com/superpronker/hz308o5bj6fs/wish/997235268 In some of the questions we use theorems from CT and rules for matrix algebra. Is it sufficient to just state that we use these rules or should we explain the rules and theorems when using them? :)]]> 2020-12-08 07:49:21 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997235268 https://padlet.com/superpronker/hz308o5bj6fs/wish/997242565 Is there a neat way of getting Matlab output into a latex format? ]]> 2020-12-08 07:52:33 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997242565 https://padlet.com/superpronker/hz308o5bj6fs/wish/997244248 Could you give some intution on how the Beta_hat values from estimator 4 (min square error of the difference between the market share and the conditional choice probabilites) should be interpreted? ]]> 2020-12-08 07:53:22 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997244248 https://padlet.com/superpronker/hz308o5bj6fs/wish/997250906 2020-12-08 07:56:32 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997250906 https://padlet.com/superpronker/hz308o5bj6fs/wish/997280914 We generally say that FE are difficult in non-linear models, while we use dummies quite often. Isn't a model with FE and dummies (e.g. over time or individuals) the same numerically as FE ? Why can we use dummies in non-linear models?]]> 2020-12-08 08:11:55 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997280914 https://padlet.com/superpronker/hz308o5bj6fs/wish/997291352 In a recent paper (Kropko & Kubinec 2020), the authors write about the interpretation of key regressors when using two-way (or more) fixed effects. They argue that controlling for more than one fixed effects actually renders the interpretation of the other variables so obscure, (by demeaning two times) that the econometrician cannot really say anything useful about estimated parameters. Thoughts? 

Cheers guys, here's the link to the article btw: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0231349]]> 2020-12-08 08:16:58 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997291352 https://padlet.com/superpronker/hz308o5bj6fs/wish/997293901
Back to the page limitation. Can you check if the page limit is also defined by characters. I ask because you typical write less than 2400 characters on 1 page, especially if you have a lot of equations as they take of a lot of space but few characters. 

Thanks!]]> 2020-12-08 08:18:17 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997293901 https://padlet.com/superpronker/hz308o5bj6fs/wish/997294084 Just to be sure I understand densities correctly. Shouldn't the nominator in the final likelihood on the slide have (y_i - x_i'beta) in parentheses. ]]> 2020-12-08 08:18:23 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997294084 https://padlet.com/superpronker/hz308o5bj6fs/wish/997305638 When should we use what strinputCovtype in our estimation?]]> 2020-12-08 08:23:59 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997305638 https://padlet.com/superpronker/hz308o5bj6fs/wish/997306443 Since you state that we should report standard errors if possible, is it common that we are not able to obtain standard error for some non-linear estimators? We obtain some estimates that are extremely large, even with fminunc arriving at 'local minimum found']]> 2020-12-08 08:24:24 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997306443 https://padlet.com/superpronker/hz308o5bj6fs/wish/997311736 Is it an acceptable argument to accept local minimum possible in project 3 due to obtaining parameter estimates that are somewhat similar to previous methods of estimating the same parameters? ]]> 2020-12-08 08:26:58 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997311736 https://padlet.com/superpronker/hz308o5bj6fs/wish/997312701 The OLS of the utility function in 2.3: 
Is the interpretaion that the beta estimates correspond to the compensating variation as a money transfer that would make the consumer indifferent between the car chosen and the alternative. 
How can we interpret a negative sign on logpr?]]> 2020-12-08 08:27:28 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997312701 https://padlet.com/superpronker/hz308o5bj6fs/wish/997316552

]]> 2020-12-08 08:29:29 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997316552 https://padlet.com/superpronker/hz308o5bj6fs/wish/997317565 To receive standard errors for ME is it always the way to go to derive the derivatives by hand and then code them up for the Delta Method. Shouldn't there be a standard thing to do in Matlab to get the derivatives for the Delta Method or is that beyond the scope ? ]]> 2020-12-08 08:30:02 UTC https://padlet.com/superpronker/hz308o5bj6fs/wish/997317565