Ok, I see what you're doing there. Plotting the difference between 2011's 15-19 year-olds and 2016's 20-24 year-olds creates the impression that over that five year period, there were 19,070 20-24 year-olds who were 15-19 in the previous census, and because there is a total of 24,345 20-24 year-olds in 2016, that must mean the 20-24 year-old population jumped by 5,275 people due to in-migration.
You're looking at the data in isolation, which leads to an incorrect conclusion. The population of 20-24 year-olds is perennially 4,000-5,000 people higher due to students arriving at universities and colleges, while local 15-19 year-olds leave for universities (or for employment, etc).
Aging the data the way you did shows this perennial in and outmigration, but it doesn't affect the overall population growth because this jump between 15-19 year-olds and 20-24 year-olds is a constant.