Page 1 of 1
Instantaneous Inflation
Posted: Thu Feb 16, 2023 3:12 pm
by AASA
Jan Eeckhout posted a draft article, "Instantaneous Inflation" dated, January 31, 2023. He uses annualized data for the prior 12 months and then uses a Kernel and Bandwidth method to weight the data and calculate annualized output for various bandwidths. Calculating the data weights for the annualized amounts seems quite laborious. Is there an automated method available on EViews? The model to be evaluated is on page 5, after various bandwidth calculations are made for the model on page 4.
Re: Instantaneous Inflation
Posted: Sun Apr 30, 2023 6:35 pm
by Elderfield.A
Hi,
You could easily write this is a subroutine/add-in. Here is some R code that estimates the weights, which took me all of 10 minutes to pull together. Would be pretty straight forward to translate this to EViews.
Adam
Code: Select all
K <- function(a, BigT){
ker <- matrix(0,BigT,1)
for(i in 1:BigT){
ker[i,] <- (BigT-(i-1))^a
}
sum_T_t_a <- sum(ker)
for(i in 1:BigT){
ker[i,] <- ker[i,]/sum_T_t_a*BigT
}
return(ker)
}
Re: Instantaneous Inflation
Posted: Mon May 08, 2023 11:15 am
by AASA
Thank you.
Being neither a wizard nor a novice on programming, I will give it a try.
Update: I was able to use Excel to complete the regression problem. 5-19-2023.
I used Excel to find the weights but am having trouble evaluating the product equation. When all exponents in the product equation are =1, the model returns the correct value. Below are the values of k(tau,4) that I found which tie-into the article. But can't seem to calculate the proper product.
tau a T k(tau,a)
0 4 12 4.0987
1 4 12 2.8940
2 4 12 1.9766
3 4 12 1.2969
4 4 12 0.8096
5 4 12 0.4746
6 4 12 0.2562
7 4 12 0.1235
8 4 12 0.0506
9 4 12 0.0160
10 4 12 0.0032
11 4 12 0.0002
Below is the table for k(tau,0)
tau a T k(tau,a)
0 0 12 1
1 0 12 1
2 0 12 1
3 0 12 1
4 0 12 1
5 0 12 1
6 0 12 1
7 0 12 1
8 0 12 1
9 0 12 1
10 0 12 1
11 0 12 1