Why not try a backtest? Pick a point in time (say, Jan 1, 2011). Pick another data point (say, Jan 1 2015). See which variables explain the rise in house prices the best?
Because past behavior is not indicative of future behavior.
If the prediction works forward, it should be backward predictive too... That's not a sufficient condition, but it is necessary to validate the predictive power, provided assumptions haven't changed in the time period being tested.