There's a simple reason why the highest quintile of almost any group of people -- under any form of measurement you wish to use (income, vital statistics, etc.) -- will increase faster than the lowest. It's because for most measurements there is no practical limitation on how high the measurement in question can go, but there is usually a practical limitation on how low it can go.
For example . . .
If you were to compare the change in life expectancy for the top 20% of the U.S. population (i.e., the 20% of the population that lives the longest) to the change in life expectancy for the lowest 20% of the U.S. population (i.e., the 20% of the population that dies earliest), you'll likely find a similar trend. This is because the distribution of the population based on age at their deaths is not a true "normal distribution" bell curve . . . people can live incrementally longer over time (someone who died at the age of 50 a century ago might live to the age of 60, 80 or even 100 today), but the lowest end of the curve is constrained at the age of 0 and cannot decrease over time.
Thanks - that makes a lot of sense.