Start with a bog-standard second-year textbook Mundell-Fleming ISLMBP model. Start in equilibrium at Y*, then hit it with a negative shock to Net eXports. The IS curve shifts left initially, at the previous equilibrium exchange rate. But the central bank is sensible, and allows the exchange rate to depreciate sufficiently to shift the IS curve right again so the new ISLMBP equilibrium is back at Y*.
Now make one small change. The shock hits Net eXports next year, not this year. But people learn about the shock this year, one year in advance. It's a news shock.
If people expect that next year's exchange rate will be (say) 10% lower than they had previously expected (before hearing the news), that means the BP curve shifts up by 10%. Which means the ISLM intersection is now below the BP curve, which means the exchange rate depreciates today. It would take a 10% depreciation of today's exchange rate (relative to the previous equilibrium) to preserve uncovered interest rate parity and bring the BP curve back down to where it was before the news shock. But that depreciation of the exchange rate shifts the IS curve rightward today. And remember, the negative shock to Net eXports hasn't hit yet, so the rightward shift in the IS curve will cause a boom. Unless the central bank takes offsetting action by tightening monetary policy.
Second year textbook macro can be quite useful.