# Testing for a substitution over time

I am trying to test what types of jobs IT employees are replacing in an industry. My data looks like this. Each of the values in the rows are percentages (I'm not sure how to put a percentage in a row in stat exchange).

\begin{array}{|c|c|c|c|} \hline Year& IT Employees & Clerical Employees & Other Employees \\ \hline 1984 & 2 & 50&48\\ \hline 1985& 4 & 47& 49\\ \hline 1986 & 7&45&48 \\ \hline \end{array}

Right now, I'm looking at all 3 pairwise correlations between ComputerEmployees, ClericalEmployees, and OtherEmployees, and seeing if the rho is significant. Of course, this ignores the fact that this is time series data. Is there a better way to do this?