statistics - Time Series Modeling of Choppy Data -

i'm trying model 10 years of monthly time series data choppy , overal has upward trend. @ first glance looks strong seasonal series, test results indicate not seasonal. pricing variable i'm trying model function of macroeconomic environment, such interest rates , yield curves. i've tryed linear ols regression (proc reg), don't goo dmodel that. i've tried autoregressive error models (proc autoreg), captures 7 lags of error term significant factors. don't want include many lag of error term in model. in addition of macroeconomic variables become insignificant when include these error lags in model.

any suggestions on modeling method/technique me model choppy data appreciated.

at past project, we've used proc arima predict future product sales based on time series of past sales: http://support.sas.com/documentation/cdl/en/etsug/60372/html/default/viewer.htm#etsug_arima_sect019.htm (note arima autoregressive model)

but joe said, statistical feedback on question, you're better of asking @ cross validated site.