[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Bootstrapped skewness-adjusted t-stat question

From   "Rajesh Tharyan" <>
To   <>
Subject   RE: st: Bootstrapped skewness-adjusted t-stat question
Date   Mon, 3 Mar 2008 13:37:14 -0000


The skewness adjustment in LBT(1999) is due to Johnson (1978).There is a
user written ado which calculates the skewness adjusted t stats for you. 

. findit johnson

I think we could incorporate this into the program to get what you want. I
am not sure if this is right but 
this is what I did. The $S_7 in the johnson ado is the skewness adjusted t

In the below program I did try to include size as follows 

bootstrap r(ratio), saving(C:\mydata, replace) reps(1000) size(10): skewt
But that didn't change anything it shows that it uses all the 74
observations. However, after I run the program and say

. return list

It shows that r(N) =  10. 

sysuse auto,clear
keep mpg 

capture program drop skewt
program define skewt, rclass
version 9
johnson mpg=0
return scalar ratio = $S_7

bootstrap r(ratio), saving(C:\mydata, replace) reps(1000): skewt

estat bootstrap, all
use C:\mydata,clear
histogram  _bs_1
centile _bs_1, centile(2.5, 97.5)


Any suggestions are valuable


-----Original Message-----
[] On Behalf Of Nuno
Sent: 02 March 2008 08:29
Subject: RE: st: Bootstrapped skewness-adjusted t-stat question

Thanks Maarten!

-----Original Message-----
[] On Behalf Of Maarten buis
Sent: 01 March 2008 23:02
Subject: Re: st: Bootstrapped skewness-adjusted t-stat question

--- Nuno <> wrote:
> I'm trying to calculate the bootstrapped skewness-adjusted t-stat 
> proposed by Lyon, Barber, and Tsai (1999), 'Improved Methods for Tests 
> of Long-Run Abnormal Stock Returns', The Journal of Finance, Vol. 54, 
> No. 1, pp. 165-201, in order to correct the skewness inherent to 
> stocks returns.

If you worry about non-normality why not go for the whole range of tried and
tested tests based on order statistics (sometimes called non-parametric
statistics). Within Stata, Roger Newson has a whole lot of packages writen
in this area, and he has also writen a number of articles on them. You can
get them at:

The approach in the paper you cite worry me a bit for two reasons (though I
did not read it very carefully) : First, it looks like you are doing some
form of bias correction using the bootstrap. The inventor of the bootstrap
warns against such bias corrections as the estimate of the bias is measured
very inacurately (Efron and Tibshirani 1993, pp. 138)). Second, the need to
bootstrap some arbitrary amount less than the number of observations in
order to get the right test indicates to me that something is not quite
right with this method. The overal feel I got was that they were pursuing a
dead-end by trying to make a t-test work for a case for which it was not
designed, while there are other tests availabel that were designed for this
Again this is just a first impression, but I would reccomend taking a good
long hard look at those non-parametric tests before continuing along this

Hope this helps,

Bradley Efron and Robert J. Tibshirani (1993) An Introduction to the
Bootstrap. Chapman & Hall/CRC.

Maarten L. Buis
Department of Social Research Methodology Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

Sent from Yahoo! Mail.
A Smarter Inbox.
*   For searches and help try:

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2019 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index