AdaemmerP
commented
5 years ago It's no bug! hp_filter() from lpirfs yields identical results as hpfilter() from mFilter. It was a mixup with some inputs. Results can be verified in R by, for example, running:
# Load libraries library(lpirfs) library(mFilter) library(testthat) library(microbenchmark)
# Use GDP series data_set <- interest_rules_var_data$GDP_gap
# hpfilter() from mFilter hp1 <- hpfilter(as.ts(data_set), freq = 1600, type = "lambda") head(hp1$trend)
# hp_filter() from lpirfs hp2 <- hp_filter(as.matrix(data_set), 1600) head(hp2[[2]])
# Verify results testthat::expect_equal(as.numeric(hp1$cycle), as.numeric(hp2[[1]]), tolerance = 0.1^10) testthat::expect_equal(as.numeric(hp1$trend), as.numeric(hp2[[2]]), tolerance = 0.1^10)
# Compare performance data_set_ts <- as.ts(data_set) data_set_mat <- as.matrix(data_set)
microbenchmark( hpfilter(data_set_ts, freq = 1600, type = "lambda"), hp_filter(data_set_mat, 1600) )
There is a bug in the hp_filter function. The value "lambda" refers to the cut-off frequency of the Hodrick-Prescott filter. To obtain values for lambda = (6, 1600, 129600), you have to type in (1, 4, 12), respectively.