Great video! I was watching you previous videos on linear models where you did partial F test to check if the variable should be included in the equation (linearHypothesis) is there any command to do the same for dynlm? And one more question if you don't mind. For the linear model you did vcovHC and vcovHAC is there any way to add these into linearhypothesis function? Would it change the results? Thank you, you are the best!
I'm not sure about dynlm but you can give it a try. As for vcovHC, you can give the to the vcov argument in linearHypothesis. This video is a bit old by the way. These days I'd probably use feols with panel.id from the fixest package instead of dynlm for many purposes, and I'd definitely always use glht from the multcomp package instead of linearHypothesis
If you want to run an ADF test with a trend and intercept you can use the ur.df function in the urca package. If you search for 'augmented dickey fuller test' in the help section this is what comes up. Alternatively I guess you could run a regression manually using the lm command as this is essentially what the ur.df function does anyway by the looks of it
Thanks a lot for your great videos I am a bit confused. The data is yearly not monthly as what I saw using print(ChinaIncome), where did the months come from when you plotted a seasonal plot? and which years are present in the graph? Thanks
I think I meant to use my simulated data for that last part, which was monthly. So what this code will do is treat the China data as though it's on a 12 year cycle of sorts.
Hi Nick! What if the ADF-test shows that the time-series/panel data is not stationary? I know in theory you have to create the derivative of a function but how do I actually implement that to the variables in R? Thank you so much!
Time series is not my primary area of expertise but I'm not sure what you mean in this context by the derivative. My intuition for non stationarity would be to difference the data, which could be done with a difference operator in a time series package, or arranging by time and subtracting a lag in dplyr
summarising the commands at the end when you were recapping the content covered was super helpful!
Thanks for your videos! Greetings from Honduras
Great video!
I was watching you previous videos on linear models where you did partial F test to check if the variable should be included in the equation (linearHypothesis) is there any command to do the same for dynlm?
And one more question if you don't mind. For the linear model you did vcovHC and vcovHAC is there any way to add these into linearhypothesis function? Would it change the results?
Thank you, you are the best!
I'm not sure about dynlm but you can give it a try. As for vcovHC, you can give the to the vcov argument in linearHypothesis.
This video is a bit old by the way. These days I'd probably use feols with panel.id from the fixest package instead of dynlm for many purposes, and I'd definitely always use glht from the multcomp package instead of linearHypothesis
If you want to run an ADF test with a trend and intercept you can use the ur.df function in the urca package. If you search for 'augmented dickey fuller test' in the help section this is what comes up. Alternatively I guess you could run a regression manually using the lm command as this is essentially what the ur.df function does anyway by the looks of it
Thanks a lot for your great videos
I am a bit confused. The data is yearly not monthly as what I saw using print(ChinaIncome), where did the months come from when you plotted a seasonal plot? and which years are present in the graph? Thanks
I think I meant to use my simulated data for that last part, which was monthly. So what this code will do is treat the China data as though it's on a 12 year cycle of sorts.
@@NickHuntingtonKlein I see. Thank you so much. I appreciate your kind help :)
Hi Nick! What if the ADF-test shows that the time-series/panel data is not stationary? I know in theory you have to create the derivative of a function but how do I actually implement that to the variables in R? Thank you so much!
Time series is not my primary area of expertise but I'm not sure what you mean in this context by the derivative. My intuition for non stationarity would be to difference the data, which could be done with a difference operator in a time series package, or arranging by time and subtracting a lag in dplyr
Hi, very helpful video. Thanks a lot!
Can you do a video about robust standard errors in Panel data? (Or link some information to it)
Kind regards.
Thanks! I'd recommend checking out my video on fixest, which can handle all your robust-SE panel data needs, at least for fixed effects modeling
It should have been...
industry