BJTRANS Procedure

@BJTrans helps choose a preliminary transformation from among none, square root and log. It graphs rescaled versions of each transformation on a single graph, and includes the log likelihoods of one lag autoregressions on the various power transformations.

@BJTRANS series start end

Parameters

series	series to analyze
start, end	range to use. By default, the range of series.

Example

open data quarterly.xls

calendar(q) 1960:1

data(format=xls,org=columns) / tbill r10 ppinsa m1nsa

* Compare log, square root, and no transformation:

@bjtrans ppinsa

Sample Graph

This is quite a smooth process so the difference isn't dramatic. However, you can see that the shorter-term fluctuations in the log are roughly the same size over the period from 1980 on, while the square root and (more so) the "none" get much more variable as the level gets larger. The criterion is much better for the log as well.