What are the chances these estimates turn out to miss the mark by such high amounts?
Not exactly answering your question, but I do know of two big Census estimate misses regarding Chicago:
The 2000 Census showed Chicago's population at 2.896 million, while in the run-up to the 2010 Census, estimates predicted a small decline to 2.853 million (so loss of just 43,000) people. When the actual numbers came out, it was...2.695 million, a loss of 201,000 people. I distinctly remember the media storm in Chicago-area media about this.
But it wasn't the first time they missed so badly regarding Chicago. In 1950, Chicago's population was 3.620 million, and in 1960, it was 3.550 million, a loss of just 70,000 people. Estimates between 1966 and 1969 predicted it would fall to close to 3.520 million, since they thought that the urban renewal projects would help stem some losses while the Great Migration Part II continued to bring in new residents, but the official numbers from the 1970 Census showed the population had dropped to 3.366 million, a loss of 184,000 people.
So I think these happen quite often, actually. I just know these two examples, however.