decrescendo ends too early

[chip]

I have a decrescendo coded like this, in 2/4 time, 2 measures -

a2~\>
a2\!

problem is the decrescendo ends at the end of the first measure, not the 
end of the second measure as it appears it should. What do I do to get 
the decrescendo to end at the end of the second measure?
--
Chip