Hi everyone,
I have a methodological question regarding the use of macro-level variables in an individual-level panel dataset for my master's thesis. When I include year fixed effects using i.year, I encounter perfect multicollinearity, which is discussed in the paper "When Macro Time Series Meets Micro Panel Data: A Clear and Present Danger" ( Ho-Chuan Huang, Xiuhua Wang, Xin Xiong)
I want to use fixed effects, but I also need to account for time while retaining the explanatory power of my macro variables. I have read that one possible solution is to demean the macro variables (i.e., subtract the yearly mean from each observation) to remove perfect multicollinearity between year dummies and macro-level data. Would this be a valid approach?
Alternatively, do you recommend any other ways to handle this issue while keeping both fixed effects and macro-level variables in the model?
Thank you in advance for your insights
I have a methodological question regarding the use of macro-level variables in an individual-level panel dataset for my master's thesis. When I include year fixed effects using i.year, I encounter perfect multicollinearity, which is discussed in the paper "When Macro Time Series Meets Micro Panel Data: A Clear and Present Danger" ( Ho-Chuan Huang, Xiuhua Wang, Xin Xiong)
I want to use fixed effects, but I also need to account for time while retaining the explanatory power of my macro variables. I have read that one possible solution is to demean the macro variables (i.e., subtract the yearly mean from each observation) to remove perfect multicollinearity between year dummies and macro-level data. Would this be a valid approach?
Alternatively, do you recommend any other ways to handle this issue while keeping both fixed effects and macro-level variables in the model?
Thank you in advance for your insights
Comment