In a way this forum isn't really about debunking statistics. It is about recognizing that statistics are really only useful to those creating them. In the case of media reporting, they are useful for making the reports exciting. We can use statistics to make decisions, but it is important for use to understand the nature of statistics to do it well.
In this case, the answer probably requires clarification of the question.
Someone else can probably set up the problem more clearly, but question is probably not "What is the divorce rate?", but rather "What is the growth rate (positive or negative) of the married population vs the growth rate (+/-) of the population in general?"
As Douglas Adams put it "42 is the answer, but what exactly was the question?"
Marriage may be in trouble but the people involved may not be, if you see what I mean.
The 50 %, as an illustrative figure, seems fair to me.
2 million in, one million out. 50% .
2 million a year and 50 million married means that, assuming the numbers are constant to make thing simple, the total number of existing marriages represents 25 years of marriage activity absent any divorces. Or 50 years approx if 2 million married each year and 1 million divorced and never re-married.
This of course discounts those who marry multiple times and who may also divorce multiple times, not necesssarily within a year of marriage.
So whereas the 50% figure may indeed apply to the marriage action statistics that does not mean that is applies to the people in the marriages. Many will have a 0% divorce, others may have, for example, 4 marriages of which 3 end in divorce. So they have either a 300% or a 75% divorce rate depending on you Point of View.
Or maybe some other number if you are reporting as an epidemiologist.