Hello,
I performed Vector error correction model using Tourist Arrivals as a dependent Variable and Real GDP, Public Road Transport operated distance and Public Railway Transport operated distance as Independent variables. Although the model was significant it did not pass the Normality test. (Image1- model1)
Then I converted two Transport variables into one using multiplication and performed the model again and this time whole model passed the normality test. (Image2- Model2)
I just want to know if can I use this Variable transformation in my model and if so could you please advise any reference for it.
Highly appreciate your comment on this.
Dear Vidusha Weesinghe
You may explore logarithmic transformation.
Please refer to
https://youtu.be/NZCSt9WkpkI?si=2sqmp1TkM6DWX33a
Hi Brajesh, The video is on interpreting regression model coefficients. I just want to know whether I could combine two time series variables to make the model normal? Do I violate any time series assumptions here?
Yes, you can, but the transformation you choose should be guided by the nature of your data and the specific requirements of your model.
Hi @Chuck, Could you please help me with any reference for this type of transformations? That will be great for my study.
You are approaching your analysis from the wrong direction. The first thing you must look at is whether your model makes economic sense.
You have four variables
y: tourism arrivals
x1: real GDP
x2: Public road ........
x3: Public railroad ........
You have three candidate models
From an economic point of view, I would think that none of these models would give a good model of tourism arrivals but that is a matter for you to decide.
If you work with 2 the marginal effect of a change in x2 is b3 + b4 x3. If b4 is negative this means that extra rail will decrease the effectiveness of road.
From an econometric point, there is no problem in including the product term. If you look up your economic textbook search for 'interaction' and you will get an explanation of interaction terms in regression.
Go through this.
https://www.bing.com/ck/a?!&&p=4abc4ed74cd145a7JmltdHM9MTcwNzYwOTYwMCZpZ3VpZD0xMzJmMWQ5ZS05NDA4LTY1ZjgtMTIyYS0wZWI1OTUxNTY0YTEmaW5zaWQ9NTE4OA&ptn=3&ver=2&hsh=3&fclid=132f1d9e-9408-65f8-122a-0eb5951564a1&psq=Can+I+transform+two+log+variables+into+one+using+multiplication+in+VECM+time+series+model%3f+&u=a1aHR0cHM6Ly9zdGF0cy5zdGFja2V4Y2hhbmdlLmNvbS9xdWVzdGlvbnMvNjM3ODcwL2Nhbi1pLXRyYW5zZm9ybS10d28tbG9nLXZhcmlhYmxlcy1pbnRvLW9uZS11c2luZy1tdWx0aXBsaWNhdGlvbi1pbi12ZWNtLXRpbWUtc2Vy&ntb=1
Yes, you can transform two log variables into one using multiplication in a VECM time series model.
- Badges
- Science topic
- Similar topics
- Agricultural Science
- Horticulture