[R-SIG-Finance] How to contribute my threshold contegration functions in R?

Matthieu Stigler matthieu.stigler at gmail.com
Sun Aug 17 22:29:35 CEST 2008


Dear members of the R-Finance list and other persons potentially concerned

This is actually a question not directly related to existing functions 
in R but rather a question on how to add new ones.

For my master thesis in economics, I had to write myself functions in R 
for threshold cointegration. I found a job after it and hoped to be able 
simultaneously to make my functions available for R , but I realized 
that it takes much more than I expected to make my raw code end-user 
suitable. I will soon finish this actual job and, before searching a 
real work, I'm thinking about spending some time to write and improve 
the functions in order to contribute them in package tsDyn (see 
http://code.google.com/p/tsdyn/wiki/ThresholdCointegration for the 
actual features).

This is why I ask you if you have an idea about how I could find a way 
to develop and improve the package?  Maybe some institute or university 
might be interested in fostering the development of a package for 
threshold cointegration and get practice in applying it, and therefore 
offer a 2-3 months job or stage place?
Cointegration analysis has become an indispensable tool/step for 
considering relationships between multivariate time series. Threshold 
cointegration extends the usual linear cointegration to cases where the 
adjustment towards long-run equilibrium does not occur after each small 
deviation but more realistically only when the deviations exceed some 
critical thresholds, thus taking transaction costs and asymmetries in 
price transmission into account .

Threshold cointegration can improve significantly the estimation and 
forecast accuracy and has been widely applied to study the relationship 
between interest rates, exchange rates, oil prices, international 
agricultural markets, and to test the purchasing power parity theory and 
the law of one price.

In spite of its wide interest in theoretical and empirical work, there 
isn't, to my knowledge, any comprehensive package or software for it, 
except some codes for Gauss and Matlab.

So I appreciate very much any idea or advice about a 2-3 month job or 
stage place to develop this package! Thank a lot!

Matthieu



More information about the R-SIG-Finance mailing list