I go back to the analogy to the stock market. By this point in time, even the great depression's effects show up as a mere blip on a graph that stretches to present day. But if you were still in the 1950's, you could greatly skew things dependent on what you picked for your base year(s).
By the same token though, perfectly valid market forecasting models could be shown to be either wildly astray or deadly accurate if you limit how far back you go and depending on how you define the base year. Using 2001 vs 2002, for instance.