## Multiple Breakpoint Test (Bai Perron) for a replication of a paper

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

hello1994
Posts: 2
Joined: Thu Sep 14, 2017 2:09 pm

### Multiple Breakpoint Test (Bai Perron) for a replication of a paper

Hi,

I am attempting to replicate a paper by Jones and Olken (ungated link here https://economics.mit.edu/files/2911)

The paper identifies the structural breaks in countries growth rates and then looks at the factors associated with them. The structural breaks are identified from this equation:

equation.png (22.51 KiB) Viewed 543 times

I presume that first the structural breaks are identified, and then a regression is applied to the breaks in order to determine the associated factors.

I downloaded PWT 6.1, and know that the authors created a growth variable by taking the difference of annual the annual real GDP per capita (Laspeyres) (rgdpl in PWT6.1). I have imitated this, and know that my results are identical to the data set used by the original authors.

From the paper I know that Algeria has a structural (down)break in growth in the year 1981, and so I wanted to replicate this break first, and then proceed to expand the procedure to the whole data set, which is where I am struggling. I know that since the publication of the paper, EViews have installed a drop down click for the implementation of the multiple breakpoint test.

This is the data I am trying to identify the break from (only Algeria).

algeriatest2.xlsx

I am estimating the equation:

growth c

and then am implementing the multiple breakpoint test with a trimming parameter of 10%; significance level of 10%; max break of 5; and the implementation method of sequential L+1 breaks vs.L) as per the paper, however I am not managing to find the structural break.

I know that I am probably making an extremely stupid mistake, any help would be invaluable!

Thanks!

hello1994
Posts: 2
Joined: Thu Sep 14, 2017 2:09 pm

### Re: Multiple Breakpoint Test (Bai Perron) for a replication of a paper

The equation in the image is:

gt = aR + et where gt is the annual growth rate in purchasing-power-parity per-capita income, aR is the mean growth rate during regime R, and et is an error term drawn from a common distribution across regimes. As before, the data is taken from the Penn World Tables v6.1.