select_lambda: Selecting lambda

In kvasilopoulos/transx: Transform Univariate Time Series

Description

Approaches to selecting lambda.

Usage

select_lambda(
  freq = c("quarterly", "annual", "monthly", "weekly"),
  type = c("rot", "ru2002")
)

Arguments

freq	[character: "quarterly"] The frequency of the dataset.
type	[character: "rot"] The methodology to select lambda.

Details

Rule of thumb is from Hodrick and Prescot (1997):

• Lambda = 100*(number of periods in a year)^2

• Annual data = 100 x 1^2 = 100

• Quarterly data = 100 x 4^2 = 1,600

• Monthly data = 100 x 12^2 = 14,400

• Weekly data = 100 x 52^2 = 270,400

• Daily data = 100 x 365^2 = 13,322,500

Ravn and Uhlig (2002) state that lambda should vary by the fourth power of the frequency observation ratio;

• Lambda = 6.25 x (number of periods in a year)^4

Thus, the rescaled default values for lambda are:

• Annual data = 1600 x 1^4 = 6.25

• Quarterly data = 1600 x 4^4= 1600

• Monthly data = 1600 x 12^4= 129,600

• Weekly data = 1600 x 12^4 = 33,177,600

References

Hodrick, R. J., & Prescott, E. C. (1997). Postwar US business cycles: an empirical investigation. Journal of Money, credit, and Banking, 1-16.

Ravn, M. O., & Uhlig, H. (2002). On adjusting the Hodrick-Prescott filter for the frequency of observations. Review of economics and statistics, 84(2), 371-376.

kvasilopoulos/transx documentation built on Jan. 26, 2021, 6:14 p.m.