I am trying to generate a 1-step-ahead forecast of a quarterly variable using a monthly variable with the
midasr package. The trouble I am having is that I can only estimate a
MIDAS model when the number of monthly observations in the sample is exactly 3 times as much the number of quarterly observations.
My question is how can I forecast in the
midasr package when the number of monthly observations is not an exact multiple of the quarterly observations (e.g. when I have a new monthly data point that I want to use to update the forecast)?
As an example, suppose I run the following code to generate a 1-step-ahead forecast when I have
(n) quarterly observations and
(3*n) monthly observations:
#first I create the quarterly and monthly variables n <- 20 qrt <- rnorm(n) mth <- rnorm(3*n) #I convert the data to time series format qrt <- ts(qrt, start = c(2009, 1), frequency = 4) mth <- ts(mth, start = c(2009, 1), frequency = 12) #now I estimate the midas model and generate a 1-step ahead forecast library(midasr) reg <- midas_r(qrt ~ mls(qrt, 1, 1) + mls(mth, 3:6, m = 3, nealmon), start = list(mth = c(1, 1, -1))) forecast(reg, newdata = list(qrt = c(NA), mth =c(NA, NA, NA)))
This code works fine. Now suppose I have a new monthly data point that I want to include, so that the new monthly data is:
nmth <- rnorm(3*n +1)
I tried running the following code to estimate the new model:
reg <- midas_r(qrt ~ mls(qrt, 1, 1) + mls(nmth, 2:7, m = 3, nealmon), start = list(mth = c(1, 1, -1))) #I now use 2 lags instead 3 with the new monthly data
However I get an error message saying:
'Error in mls(nmth, 2:7, m = 3, nealmon) : Incomplete high frequency data'
I could not find anything online on how to deal with this problem. Any help is greatly appreciated.