The RATS Software Forum

@LPUNIT implements the Lumsdaine-Papell unit root test, allowing for two breaks in the intercept, the trend or both at unknown locations. Lumsdaine and Papell(1997), "Multiple trend breaks and the unit root hypothesis", Review of Economics and Statistics, vol 79, 212-218.

Detailed description

Dear Tom,

I've tried working with the LPUnit using the following options:
BREAK=TREND
NBREAKS=3
GRAPH

As i am uusing the entire observation i didn't include the start and end dates.

I tested for US and GERMANY Series

So the command is:
LPUnit(BREAK=TREND,NBREAK=3,GRAPH) US

After letting it run for several hour and no results, I had to cancel.

The funny thing is that when i tried the same test but with NBREAKS=2 and a smaller series the test took only 30 mins approximately..

So can you tell me if the error is in the NBREAKS or the size of the series.

I have attached both data files the first one is the longest and the second one is the shortest.

Many Thanks,

Bachar

If 2 breaks took 30 minutes, it's not a "funny thing" that 3 takes a very, very long time. That's the nature of the calculation. Given the size of your data set, a 3 break test will probably take somewhere around two months, give or take a few weeks. You might want to read Section 11.3 of the RATS User's Guide.

If I reduce it to 2 breaks given the size of my longest series, how long will it take?

Read section 11.3 of the User's Guide. It explains how the complexity of the calculation changes with data set size and number of breaks.