to_add_to_pkg/Recdev autocorrelation high vs low.r

In CCSBT/sbtr: sbtr

#########################################################
#Is autocorrelation bigger for positive recruitment deviations
#or for negative recruitment deviations? 
#Go through the data and divide into high rec devs (1960-2008)
#and low rec devs. Then get the rec dev the following year.
#For these two groups (high and low) calculate the autocorrelation
#to see if this is higher or lower for high recruitment. 
#Current version created by Trevor Branch tbranch@gmail.com
#starting 27 June 2014
#########################################################