Skip to contents

Beveridge-Nelson Filter

bnf.Rd

Based on Replication Files for Kamber, Morley & Wong R conversion of some of the MATLAB codes originally written by Ben Wong

Usage

bnf(
  y,
  delta_select = 2,
  fixed_delta = 0.01,
  p = 12,
  d0 = 0.01,
  dt = 5e-04,
  demean = c("nd", "sm", "pm", "dm"),
  iterative = 100,
  dynamic_bands = TRUE,
  adjusted_bands = FALSE,
  outliers = c(293, 294),
  window = 40,
  ib = TRUE,
  ...
)

Arguments

y

vector, eg object of type ts

delta_select

integer, set to 0 if use fixed delta, 1 if max amp-to-noise, 2 if min var(trend shocks)

fixed_delta

integer, set a fixed delta to be used if delta_select=0, or for delta

p

integer, is lag order for estimation

d0

integer, min_delta: start point of grid search for delta if delta_select!=0

dt

integer, delta increment

demean

character, one of "nd", "sm", "dm", or "pm", where "nd" = no drift, "sm" = sample mean, "dm" = dynamic demeaning, "pm" = structural breaks, set as 'breaks = c(100, 237)'

iterative

integer, set to >1 for max number of iterations for iterative dynamic demeaning

dynamic_bands

logical, set to TRUE for dynamic error bands, FALSE for fixed standard error bands

adjusted_bands

logical, set to TRUE to adjusts for outlier observations when calculating error bands,

outliers

set outliers as 'outliers = c(293, 294)'

window

rolling window length for dynamic demeaning and/or dynamic error bands (e.g., 40 is 10 years for quarterly data)

ib

logical, set to FALSE if no iterative backcasting as in KMW2018 (just unconditional mean), set to TRUE if iterative backcasting

...

passed into error bands and piecewise_demean function

Value

List of class "bnf" containing:

  • call

  • y

  • cycle

  • trend

  • cycle_se: standard error

  • delta

  • demean_method

  • iterative

  • cycle_ci

  • cycle_adjusted_se

  • cycle_ci_adjusted

  • iterations

References

  • Kamber G, Morley J, Wong B (2018). “Intuitive and Reliable Estimates of the Output Gap from a Beveridge-Nelson Filter.” The Review of Economics and Statistics, 100(3), 550-566. ISSN 0034-6535,.

  • Kamber G, Morley J, Wong B (2025). “Trend-cycle decomposition in the presence of large shocks.” Journal of Economic Dynamics and Control, 173, 105066. ISSN 0165-1889.

Examples

data(usdata)
y <- transform_series(y = usdata$GDPC1, take_log = TRUE, pcode = "p1")
bnfOutput <- bnf(as.vector(y),
        delta_select = 2,
        demean = "dm",
        iterative = 100,
        dynamic_bands = TRUE,
        adjusted_bands = TRUE,
        outliers = c(293, 294),
        window = 40,
        ib = TRUE)